Publications resulting from research conducted using Delta appear here. Check back to see how the list of exciting discoveries made using Delta grows.
If you have a publication that should be listed here and isn’t, please share your success with us!
5005740
XHSH9DGT
1
nature
50
default
1693
https://delta.ncsa.illinois.edu/wp-content/plugins/zotpress/
%7B%22status%22%3A%22success%22%2C%22updateneeded%22%3Afalse%2C%22instance%22%3Afalse%2C%22meta%22%3A%7B%22request_last%22%3A450%2C%22request_next%22%3A50%2C%22used_cache%22%3Atrue%7D%2C%22data%22%3A%5B%7B%22key%22%3A%22SSDMAYMZ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Yu%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BYu%2C%20H.%2C%20Ahuja%2C%20K.%20K.%2C%20Sankar%2C%20L.%20N.%20%26amp%3B%20Bryngelson%2C%20S.%20H.%20Transmission%20of%20High-Amplitude%20Sound%20through%20Leakages%20of%20Ill-fitting%20Earplugs.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.16355%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.16355%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Transmission%20of%20High-Amplitude%20Sound%20through%20Leakages%20of%20Ill-fitting%20Earplugs%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Haocheng%22%2C%22lastName%22%3A%22Yu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Krishan%20K.%22%2C%22lastName%22%3A%22Ahuja%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lakshmi%20N.%22%2C%22lastName%22%3A%22Sankar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Spencer%20H.%22%2C%22lastName%22%3A%22Bryngelson%22%7D%5D%2C%22abstractNote%22%3A%22High%20sound%20pressure%20levels%20%28SPL%29%20pose%20notable%20risks%20in%20loud%20environments%2C%20particularly%20due%20to%20noise-induced%20hearing%20loss.%20Ill-fitting%20earplugs%20often%20lead%20to%20sound%20leakage%2C%20a%20phenomenon%20this%20study%20seeks%20to%20investigate.%20To%20validate%20our%20methodology%2C%20we%20first%20obtained%20computational%20and%20experimental%20acoustic%20transmission%20data%20for%20stand-alone%20slit%20resonators%20and%20orifices%2C%20for%20which%20extensive%20published%20data%20are%20readily%20available%20for%20comparison.%20We%20then%20examined%20the%20frequency-dependent%20acoustic%20power%20absorption%20coefficient%20and%20transmission%20loss%20%28TL%29%20across%20various%20leakage%20geometries%2C%20modeled%20using%20different%20orifice%20diameters.%20Experimental%20approaches%20spanned%20a%20frequency%20range%20of%201--5%20kHz%20under%20SPL%20conditions%20of%20120--150%20dB.%20Key%20findings%20reveal%20that%20unsealed%20silicone%20rubber%20earplugs%20demonstrate%20an%20average%20TL%20reduction%20of%20approximately%2018%20dB%20at%20an%20overall%20incident%20SPL%20%28OISPL%29%20of%20120%20dB.%20Direct%20numerical%20simulations%20further%20highlight%20SPL-dependent%20acoustic%20dissipation%20mechanisms%2C%20showing%20the%20conversion%20of%20acoustic%20energy%20into%20vorticity%20in%20ill-fitting%20earplug%20models%20at%20an%20OISPL%20of%20150%20dB.%20These%20results%20highlight%20the%20role%20of%20earplug%20design%20for%20high-sound-pressure-level%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.16355%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.16355%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-02-26T15%3A05%3A36Z%22%7D%7D%2C%7B%22key%22%3A%22QPNR92SN%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Xiao%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BXiao%2C%20X.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Prompt-based%20Adaptation%20in%20Large-scale%20Vision%20Models%3A%20A%20Survey.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.13219%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.13219%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Prompt-based%20Adaptation%20in%20Large-scale%20Vision%20Models%3A%20A%20Survey%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xi%22%2C%22lastName%22%3A%22Xiao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yunbei%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lin%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yiyang%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiaoying%22%2C%22lastName%22%3A%22Liao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zheda%22%2C%22lastName%22%3A%22Mai%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xingjian%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiao%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hao%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jihun%22%2C%22lastName%22%3A%22Hamm%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xue%22%2C%22lastName%22%3A%22Lin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Min%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qifan%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tianyang%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cheng%22%2C%22lastName%22%3A%22Han%22%7D%5D%2C%22abstractNote%22%3A%22In%20computer%20vision%2C%20Visual%20Prompting%20%28VP%29%20and%20Visual%20Prompt%20Tuning%20%28VPT%29%20have%20recently%20emerged%20as%20lightweight%20and%20effective%20alternatives%20to%20full%20fine-tuning%20for%20adapting%20large-scale%20vision%20models%20within%20the%20%60%60pretrain-then-finetune%26%23039%3B%26%23039%3B%20paradigm.%20However%2C%20despite%20rapid%20progress%2C%20their%20conceptual%20boundaries%20remain%20blurred%2C%20as%20VP%20and%20VPT%20are%20frequently%20used%20interchangeably%20in%20current%20research%2C%20reflecting%20a%20lack%20of%20systematic%20distinction%20between%20these%20techniques%20and%20their%20respective%20applications.%20In%20this%20survey%2C%20we%20revisit%20the%20designs%20of%20VP%20and%20VPT%20from%20first%20principles%2C%20and%20conceptualize%20them%20within%20a%20unified%20framework%20termed%20Prompt-based%20Adaptation%20%28PA%29.%20We%20provide%20a%20taxonomy%20that%20categorizes%20existing%20methods%20into%20learnable%2C%20generative%2C%20and%20non-learnable%20prompts%2C%20and%20further%20organizes%20them%20by%20injection%20granularity%20--%20pixel-level%20and%20token-level.%20Beyond%20the%20core%20methodologies%2C%20we%20examine%20PA%26%23039%3Bs%20integrations%20across%20diverse%20domains%2C%20including%20medical%20imaging%2C%203D%20point%20clouds%2C%20and%20vision-language%20tasks%2C%20as%20well%20as%20its%20role%20in%20test-time%20adaptation%20and%20trustworthy%20AI.%20We%20also%20summarize%20current%20benchmarks%20and%20identify%20key%20challenges%20and%20future%20directions.%20To%20the%20best%20of%20our%20knowledge%2C%20we%20are%20the%20first%20comprehensive%20survey%20dedicated%20to%20PA%26%23039%3Bs%20methodologies%20and%20applications%20in%20light%20of%20their%20distinct%20characteristics.%20Our%20survey%20aims%20to%20provide%20a%20clear%20roadmap%20for%20researchers%20and%20practitioners%20in%20all%20area%20to%20understand%20and%20explore%20the%20evolving%20landscape%20of%20PA-related%20research.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.13219%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.13219%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-02-26T15%3A02%3A26Z%22%7D%7D%2C%7B%22key%22%3A%22SN8ZKEPH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Maurel-Oujia%20and%20Maeda%22%2C%22parsedDate%22%3A%222025-10-15%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMaurel-Oujia%2C%20T.%20%26amp%3B%20Maeda%2C%20K.%20Subcontinuum%20structures%20of%20reactive%20shock%20waves%20in%20gaseous%20H%202%20%5C%2F%20O%202%20mixtures.%20%26lt%3Bi%26gt%3BPhys.%20Rev.%20Fluids%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B10%26lt%3B%5C%2Fb%26gt%3B%2C%20103201%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Subcontinuum%20structures%20of%20reactive%20shock%20waves%20in%20gaseous%20H%202%20%5C%2F%20O%202%20mixtures%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Thibault%22%2C%22lastName%22%3A%22Maurel-Oujia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kazuki%22%2C%22lastName%22%3A%22Maeda%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22date%22%3A%222025-10-15%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1103%5C%2Fy7lz-563m%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Flink.aps.org%5C%2Fdoi%5C%2F10.1103%5C%2Fy7lz-563m%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222469-990X%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T15%3A37%3A31Z%22%7D%7D%2C%7B%22key%22%3A%22E84RBSDZ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Monea%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMonea%2C%20G.%2C%20Feldman%2C%20Y.%2C%20Padmanabhan%2C%20S.%2C%20Brantley%2C%20K.%20%26amp%3B%20Artzi%2C%20Y.%20Breadcrumbs%20Reasoning%3A%20Memory-Efficient%20Reasoning%20with%20Compression%20Beacons.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.13797%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.13797%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Breadcrumbs%20Reasoning%3A%20Memory-Efficient%20Reasoning%20with%20Compression%20Beacons%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Giovanni%22%2C%22lastName%22%3A%22Monea%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yair%22%2C%22lastName%22%3A%22Feldman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shankar%22%2C%22lastName%22%3A%22Padmanabhan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kiant%5Cu00e9%22%2C%22lastName%22%3A%22Brantley%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yoav%22%2C%22lastName%22%3A%22Artzi%22%7D%5D%2C%22abstractNote%22%3A%22The%20scalability%20of%20large%20language%20models%20for%20long-context%20reasoning%20is%20severely%20constrained%20by%20the%20linear%20growth%20of%20their%20Transformer%20key-value%20cache%2C%20which%20incurs%20significant%20memory%20and%20computational%20costs.%20We%20posit%20that%20as%20a%20model%20generates%20reasoning%20tokens%2C%20the%20informational%20value%20of%20past%20generated%20tokens%20diminishes%2C%20creating%20an%20opportunity%20for%20compression.%20In%20this%20work%2C%20we%20propose%20to%20periodically%20compress%20the%20generation%20KV%20cache%20with%20a%20learned%2C%20special-purpose%20token%20and%20evict%20compressed%20entries.%20We%20train%20the%20model%20to%20perform%20this%20compression%20via%20a%20modified%20joint%20distillation%20and%20reinforcement%20learning%20%28RL%29%20framework.%20Our%20training%20method%20minimizes%20overhead%20over%20the%20conventional%20RL%20process%2C%20as%20it%20leverages%20RL%20outputs%20for%20distillation.%20Empirically%2C%20our%20method%20achieves%20a%20superior%20memory-accuracy%20Pareto%20frontier%20compared%20to%20both%20the%20model%20without%20cache%20compression%20and%20training-free%20compression%20techniques.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.13797%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.13797%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T15%3A34%3A40Z%22%7D%7D%2C%7B%22key%22%3A%22NZHF6RA3%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Xia%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BXia%2C%20J.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Inpainting%20the%20Neural%20Picture%3A%20Inferring%20Unrecorded%20Brain%20Area%20Dynamics%20from%20Multi-Animal%20Datasets.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.11924%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.11924%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Inpainting%20the%20Neural%20Picture%3A%20Inferring%20Unrecorded%20Brain%20Area%20Dynamics%20from%20Multi-Animal%20Datasets%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ji%22%2C%22lastName%22%3A%22Xia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yizi%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shuqi%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Genevera%20I.%22%2C%22lastName%22%3A%22Allen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Liam%22%2C%22lastName%22%3A%22Paninski%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cole%20Lincoln%22%2C%22lastName%22%3A%22Hurwitz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kenneth%20D.%22%2C%22lastName%22%3A%22Miller%22%7D%5D%2C%22abstractNote%22%3A%22Characterizing%20interactions%20between%20brain%20areas%20is%20a%20fundamental%20goal%20of%20systems%20neuroscience.%20While%20such%20analyses%20are%20possible%20when%20areas%20are%20recorded%20simultaneously%2C%20it%20is%20rare%20to%20observe%20all%20combinations%20of%20areas%20of%20interest%20within%20a%20single%20animal%20or%20recording%20session.%20How%20can%20we%20leverage%20multi-animal%20datasets%20to%20better%20understand%20multi-area%20interactions%3F%20Building%20on%20recent%20progress%20in%20large-scale%2C%20multi-animal%20models%2C%20we%20introduce%20NeuroPaint%2C%20a%20masked%20autoencoding%20approach%20for%20inferring%20the%20dynamics%20of%20unrecorded%20brain%20areas.%20By%20training%20across%20animals%20with%20overlapping%20subsets%20of%20recorded%20areas%2C%20NeuroPaint%20learns%20to%20reconstruct%20activity%20in%20missing%20areas%20based%20on%20shared%20structure%20across%20individuals.%20We%20train%20and%20evaluate%20our%20approach%20on%20synthetic%20data%20and%20two%20multi-animal%2C%20multi-area%20Neuropixels%20datasets.%20Our%20results%20demonstrate%20that%20models%20trained%20across%20animals%20with%20partial%20observations%20can%20successfully%20in-paint%20the%20dynamics%20of%20unrecorded%20areas%2C%20enabling%20multi-area%20analyses%20that%20transcend%20the%20limitations%20of%20any%20single%20experiment.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.11924%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.11924%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T15%3A28%3A31Z%22%7D%7D%2C%7B%22key%22%3A%22P6J7IMIT%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Ahmed%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BAhmed%2C%20M.%20T.%2C%20Ahmed%2C%20F.%20%26amp%3B%20Li%2C%20J.%20Enhancement%20of%20diffusivity%20and%20plastic%20deformation%20in%20ultrasound-assisted%20cold%20spray%20of%20tungsten%3A%20a%20molecular%20dynamics%20study.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.09803%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.09803%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Enhancement%20of%20diffusivity%20and%20plastic%20deformation%20in%20ultrasound-assisted%20cold%20spray%20of%20tungsten%3A%20a%20molecular%20dynamics%20study%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Md%20Tusher%22%2C%22lastName%22%3A%22Ahmed%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Farid%22%2C%22lastName%22%3A%22Ahmed%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jianzhi%22%2C%22lastName%22%3A%22Li%22%7D%5D%2C%22abstractNote%22%3A%22Tungsten%20%28%24W%24%29%20is%20widely%20valued%20for%20its%20exceptional%20thermal%20stability%2C%20mechanical%20strength%2C%20and%20corrosion%20resistance%2C%20making%20it%20an%20ideal%20candidate%20for%20high-performance%20military%20and%20aerospace%20applications.%20However%2C%20its%20high%20melting%20point%20and%20inherent%20brittleness%20pose%20significant%20challenges%20for%20processing%20%24W%24%20using%20additive%20manufacturing%20%28AM%29.%20Cold%20spray%20%28CS%29%2C%20a%20solid-state%20AM%20process%20that%20relies%20on%20high-velocity%20particle%20impact%20and%20plastic%20deformation%2C%20offers%20a%20promising%20alternative.%20In%20this%20study%2C%20we%20employ%20atomistic%20simulations%20to%20investigate%20the%20feasibility%20of%20CS%20for%20tungsten.%20We%20show%20that%20ultrasound%20perturbation%20can%20significantly%20enhance%20the%20self-diffusivity%20and%20plastic%20deformation%20of%20%24W%24%20compared%20to%20the%20negligible%20diffusion%20and%20plastic%20deformation%20observed%20in%20non-ultrasound-assisted%20CS%20of%20%24W%24.%20For%20different%20impact%20velocities%2C%20particle%20sizes%2C%20and%20ultrasound%20parameters%2C%20we%20demonstrate%20that%20ultrasound-assisted%20viscoplasticity%20enhances%20self-diffusivity%20by%20inhibiting%20grain%20boundaries%20and%20incorporating%20softening%20in%20%24W%24.%20Moreover%2C%20we%20found%20that%20this%20enhanced%20diffusion%20in%20ultrasound-assisted%20%24W%24%20can%20be%20exploited%20to%20promote%20interdiffusion%20at%20the%20particle-substrate%20interface%2C%20enabling%20in%20situ%20alloy%20formation.%20Through%20the%20formation%20of%20an%20equimolar%20%24V%24-%24W%24%20alloy%20on%20a%20%24W%24%20substrate%20using%20ultrasound-assisted%20CS%20simulations%2C%20we%20observed%20distinct%20mechanical%20properties%20and%20a%20reduced%20dislocation%20density%20in%20the%20deposited%20coating%20compared%20to%20a%20pure%20tungsten%20substrate.%20These%20results%20highlight%20the%20potential%20of%20ultrasound-assisted%20CS%20as%20a%20viable%20approach%20for%20manufacturing%20uniform%20coatings%20and%20engineered%20alloys%2C%20addressing%20key%20limitations%20in%20the%20AM%20of%20refractory%20metals.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.09803%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.09803%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T15%3A24%3A34Z%22%7D%7D%2C%7B%22key%22%3A%22U8L2X9UU%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Dhruv%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BDhruv%2C%20V.%2C%20Prather%2C%20B.%2C%20Chandra%2C%20M.%2C%20Joshi%2C%20A.%20V.%20%26amp%3B%20Gammie%2C%20C.%20F.%20Electromagnetic%20Observables%20of%20Weakly%20Collisional%20Black%20Hole%20Accretion.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.11365%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.11365%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Electromagnetic%20Observables%20of%20Weakly%20Collisional%20Black%20Hole%20Accretion%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Vedant%22%2C%22lastName%22%3A%22Dhruv%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ben%22%2C%22lastName%22%3A%22Prather%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mani%22%2C%22lastName%22%3A%22Chandra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Abhishek%20V.%22%2C%22lastName%22%3A%22Joshi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Charles%20F.%22%2C%22lastName%22%3A%22Gammie%22%7D%5D%2C%22abstractNote%22%3A%22The%20black%20holes%20in%20the%20Event%20Horizon%20Telescope%20sources%20Messier%2087%2A%20and%20Sagittarius%20A%2A%20%28SgrA%2A%29%20are%20embedded%20in%20a%20hot%2C%20collisionless%20plasma%20that%20is%20fully%20described%20in%20kinetic%20theory%20yet%20is%20usually%20modeled%20as%20an%20ideal%2C%20magnetized%20fluid.%20In%20this%20Letter%2C%20we%20present%20results%20from%20a%20new%20set%20of%20weakly%20collisional%20fluid%20simulations%20in%20which%20leading%20order%20kinetic%20effects%20are%20modeled%20as%20viscosity%20and%20heat%20conduction.%20Consistent%20with%20earlier%2C%20lower-resolution%20studies%2C%20we%20find%20that%20overall%20flow%20dynamics%20remain%20very%20similar%20between%20ideal%20and%20non-ideal%20models.%20For%20the%20first%20time%2C%20we%20synthesize%20images%20and%20spectra%20of%20SgrA%2A%20from%20weakly%20collisional%20models%20--%20assuming%20an%20isotropic%2C%20thermal%20population%20of%20electrons%20--%20and%20find%20that%20these%20remain%20largely%20indistinguishable%20from%20ideal%20fluid%20predictions.%20However%2C%20most%20weakly%20collisional%20models%20exhibit%20lower%20light%20curve%20variability%2C%20with%20all%20magnetically%20dominated%20models%20showing%20a%20small%20but%20systematic%20decrease%20in%20variability.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.11365%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.11365%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A55%3A53Z%22%7D%7D%2C%7B%22key%22%3A%224ZMNJNUV%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Halychanskyi%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHalychanskyi%2C%20Y.%2C%20Churchwell%2C%20C.%2C%20Wen%2C%20Y.%20%26amp%3B%20Kindratenko%2C%20V.%20FAC-FACodec%3A%20Controllable%20Zero-Shot%20Foreign%20Accent%20Conversion%20with%20Factorized%20Speech%20Codec.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.10785%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.10785%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22FAC-FACodec%3A%20Controllable%20Zero-Shot%20Foreign%20Accent%20Conversion%20with%20Factorized%20Speech%20Codec%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yurii%22%2C%22lastName%22%3A%22Halychanskyi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cameron%22%2C%22lastName%22%3A%22Churchwell%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yutong%22%2C%22lastName%22%3A%22Wen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Volodymyr%22%2C%22lastName%22%3A%22Kindratenko%22%7D%5D%2C%22abstractNote%22%3A%22Previous%20accent%20conversion%20%28AC%29%20methods%2C%20including%20foreign%20accent%20conversion%20%28FAC%29%2C%20lack%20explicit%20control%20over%20the%20degree%20of%20modification.%20Because%20accent%20modification%20can%20alter%20the%20perceived%20speaker%20identity%2C%20balancing%20conversion%20strength%20and%20identity%20preservation%20is%20crucial.%20We%20present%20an%20AC%20framework%20that%20provides%20an%20explicit%2C%20user-controllable%20parameter%20for%20accent%20modification.%20The%20method%20targets%20pronunciation%20while%20preserving%20suprasegmental%20cues%20such%20as%20intonation%20and%20phoneme%20durations.%20Results%20show%20performance%20comparable%20to%20recent%20AC%20systems%2C%20stronger%20preservation%20of%20speaker%20identity%2C%20and%20unique%20support%20for%20controllable%20accent%20conversion.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.10785%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.10785%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A49%3A17Z%22%7D%7D%2C%7B%22key%22%3A%229Y3NILUS%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Guan%20and%20Balakrishna%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BGuan%2C%20H.%20%26amp%3B%20Balakrishna%2C%20A.%20R.%20CuPyMag%3A%20GPU-Accelerated%20Finite-Element%20Micromagnetics%20with%20Magnetostriction.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.09812%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.09812%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22CuPyMag%3A%20GPU-Accelerated%20Finite-Element%20Micromagnetics%20with%20Magnetostriction%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hongyi%22%2C%22lastName%22%3A%22Guan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ananya%20Renuka%22%2C%22lastName%22%3A%22Balakrishna%22%7D%5D%2C%22abstractNote%22%3A%22We%20introduce%20CuPyMag%2C%20an%20open-source%2C%20Python-based%20framework%20for%20large-scale%20micromagnetic%20simulations%20with%20magnetostriction.%20CuPyMag%20solves%20micromagnetics%20with%20finite%20elements%20in%20a%20GPU-resident%20workflow%20in%20which%20key%20operations%2C%20such%20as%20right-hand-side%20assembly%2C%20spatial%20derivatives%2C%20and%20volume%20averages%2C%20are%20tensorized%20using%20CuPy%26%23039%3Bs%20BLAS-accelerated%20backend.%20Benchmark%20tests%20show%20that%20the%20GPU%20solvers%20in%20CuPyMag%20achieve%20a%20speedup%20of%20up%20to%20two%20orders%20of%20magnitude%20compared%20to%20the%20CPU%20codes.%20Its%20runtime%20grows%20linearly%5C%2Fsublinearly%20with%20problem%20size%2C%20demonstrating%20high%20efficiency.%20Additionally%2C%20CuPyMag%20uses%20the%20Gauss-Seidel%20projection%20method%20for%20time%20integration%2C%20which%20not%20only%20allows%20stable%20time%20steps%20%28up%20to%2011%20ps%29%20but%20also%20solves%20each%20governing%20equation%20with%20only%201-3%20conjugate-gradient%20iterations%20without%20preconditioning.%20CuPyMag%20accounts%20for%20magnetoelastic%20coupling%20and%20far-field%20effects%20arising%20from%20the%20boundary%20of%20the%20magnetic%20body%2C%20both%20of%20which%20play%20an%20important%20role%20in%20magnetization%20reversal%20in%20the%20presence%20of%20local%20defects.%20CuPyMag%20solves%20these%20computationally%20intensive%20multiphysics%20simulations%20with%20a%20high-resolution%20mesh%20%28up%20to%203M%20nodes%29%20in%20under%20three%20hours%20on%20an%20NVIDIA%20H200%20GPU.%20This%20acceleration%20enables%20micromagnetic%20simulations%20with%20non-trivial%20defect%20geometries%20and%20resolves%20nanoscale%20magnetic%20structures.%20It%20expands%20the%20scope%20of%20micromagnetic%20simulations%20towards%20realistic%2C%20large-scale%20problems%20that%20can%20guide%20experiments.%20More%20broadly%2C%20CuPyMag%20is%20developed%20using%20widely%20adopted%20Python%20libraries%2C%20which%20provide%20cross-platform%20compatibility%2C%20ease%20of%20installation%2C%20and%20accessibility%20for%20adaptations%20to%20diverse%20applications.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.09812%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.09812%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A36%3A58Z%22%7D%7D%2C%7B%22key%22%3A%22TXQENHKH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Liu%20et%20al.%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLiu%2C%20P.%2C%20Rhee%2C%20S.%2C%20Wilkins%2C%20M.%20%26amp%3B%20Dinda%2C%20P.%20Parameterized%20Algorithms%20and%20Parameter%20Selection%20for%20Fast%20GPU-GPU%20Collective%20Communication.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fscholar.google.com%5C%2Fcitations%3Fview_op%3Dview_citation%26amp%3Bhl%3Den%26amp%3Buser%3DwWZq-M0AAAAJ%26amp%3Bcitation_for_view%3DwWZq-M0AAAAJ%3AzYLM7Y9cAGgC%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fscholar.google.com%5C%2Fcitations%3Fview_op%3Dview_citation%26amp%3Bhl%3Den%26amp%3Buser%3DwWZq-M0AAAAJ%26amp%3Bcitation_for_view%3DwWZq-M0AAAAJ%3AzYLM7Y9cAGgC%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Parameterized%20Algorithms%20and%20Parameter%20Selection%20for%20Fast%20GPU-GPU%20Collective%20Communication%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Peizhi%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sean%22%2C%22lastName%22%3A%22Rhee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%22%2C%22lastName%22%3A%22Wilkins%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Peter%22%2C%22lastName%22%3A%22Dinda%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%22%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fscholar.google.com%5C%2Fcitations%3Fview_op%3Dview_citation%26hl%3Den%26user%3DwWZq-M0AAAAJ%26citation_for_view%3DwWZq-M0AAAAJ%3AzYLM7Y9cAGgC%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A32%3A24Z%22%7D%7D%2C%7B%22key%22%3A%2233UN2GBQ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Hayes%20and%20Krishnan%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHayes%2C%20T.%20L.%20%26amp%3B%20Krishnan%2C%20G.%20P.%20Quantifying%20the%20Role%20of%20OpenFold%20Components%20in%20Protein%20Structure%20Prediction.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.14781%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.14781%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Quantifying%20the%20Role%20of%20OpenFold%20Components%20in%20Protein%20Structure%20Prediction%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tyler%20L.%22%2C%22lastName%22%3A%22Hayes%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Giri%20P.%22%2C%22lastName%22%3A%22Krishnan%22%7D%5D%2C%22abstractNote%22%3A%22Models%20such%20as%20AlphaFold2%20and%20OpenFold%20have%20transformed%20protein%20structure%20prediction%2C%20yet%20their%20inner%20workings%20remain%20poorly%20understood.%20We%20present%20a%20methodology%20to%20systematically%20evaluate%20the%20contribution%20of%20individual%20OpenFold%20components%20to%20structure%20prediction%20accuracy.%20We%20identify%20several%20components%20that%20are%20critical%20for%20most%20proteins%2C%20while%20others%20vary%20in%20importance%20across%20proteins.%20We%20further%20show%20that%20the%20contribution%20of%20several%20components%20is%20correlated%20with%20protein%20length.%20These%20findings%20provide%20insight%20into%20how%20OpenFold%20achieves%20accurate%20predictions%20and%20highlight%20directions%20for%20interpreting%20protein%20prediction%20networks%20more%20broadly.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2511.14781%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2511.14781%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A18%3A43Z%22%7D%7D%2C%7B%22key%22%3A%22PYJXLVUZ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Liu%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLiu%2C%20L.%2C%20Qin%2C%20F.%2C%20Legunsen%2C%20O.%20%26amp%3B%20d%26%23x2019%3BAmorim%2C%20M.%20Bug%20Histories%20as%20Sources%20of%20Compiler%20Fuzzing%20Mutators.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.07834%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.07834%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Bug%20Histories%20as%20Sources%20of%20Compiler%20Fuzzing%20Mutators%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lingjun%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Feiran%22%2C%22lastName%22%3A%22Qin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Owolabi%22%2C%22lastName%22%3A%22Legunsen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marcelo%22%2C%22lastName%22%3A%22d%27Amorim%22%7D%5D%2C%22abstractNote%22%3A%22Bugs%20in%20compilers%2C%20which%20are%20critical%20infrastructure%20today%2C%20can%20have%20outsized%20negative%20impacts.%20Mutational%20fuzzers%20aid%20compiler%20bug%20detection%20by%20systematically%20mutating%20compiler%20inputs%2C%20i.e.%2C%20programs.%20Their%20effectiveness%20depends%20on%20the%20quality%20of%20the%20mutators%20used.%20Yet%2C%20no%20prior%20work%20used%20compiler%20bug%20histories%20as%20a%20source%20of%20mutators.%20We%20propose%20IssueMut%2C%20the%20first%20approach%20for%20extracting%20compiler%20fuzzing%20mutators%20from%20bug%20histories.%20Our%20insight%20is%20that%20bug%20reports%20contain%20hints%20about%20program%20elements%20that%20induced%20compiler%20bugs%3B%20they%20can%20guide%20fuzzers%20towards%20similar%20bugs.%20IssueMut%20uses%20an%20automated%20method%20to%20mine%20mutators%20from%20bug%20reports%20and%20retrofit%20such%20mutators%20into%20existing%20mutational%20compiler%20fuzzers.%20Using%20IssueMut%2C%20we%20mine%20587%20mutators%20from%201760%20GCC%20and%20LLVM%20bug%20reports.%20Then%2C%20we%20run%20IssueMut%20on%20these%20compilers%2C%20with%20all%20their%20test%20inputs%20as%20seed%20corpora.%20We%20find%20that%20%26quot%3Bbug%20history%26quot%3B%20mutators%20are%20effective%3A%20they%20find%20new%20bugs%20that%20a%20state-of-the-art%20mutational%20compiler%20fuzzer%20misses-28%20in%20GCC%20and%2037%20in%20LLVM.%20Of%20these%2C%2060%20were%20confirmed%20or%20fixed%2C%20validating%20our%20idea%20that%20bug%20histories%20have%20rich%20information%20that%20compiler%20fuzzers%20should%20leverage.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.07834%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.07834%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-12T14%3A14%3A18Z%22%7D%7D%2C%7B%22key%22%3A%229C9SQ7GI%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Alatrush%20et%20al.%22%2C%22parsedDate%22%3A%222025-09%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BAlatrush%2C%20N.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Advancing%20Active%20Learning%20with%20Ensemble%20Strategies.%20in%20%28Varna%2C%20Bulgaria%2C%202025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Advancing%20Active%20Learning%20with%20Ensemble%20Strategies%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Naif%22%2C%22lastName%22%3A%22Alatrush%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sultan%22%2C%22lastName%22%3A%22Alsarra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Afraa%22%2C%22lastName%22%3A%22Alshammari%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Luay%22%2C%22lastName%22%3A%22Abdeljaber%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Niamat%22%2C%22lastName%22%3A%22Zawad%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Latifur%22%2C%22lastName%22%3A%22Khan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Patrick%20T%22%2C%22lastName%22%3A%22Brandt%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Javier%22%2C%22lastName%22%3A%22Osorio%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Vito%20J%22%2C%22lastName%22%3A%22D%27Orazio%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%22%22%2C%22conferenceName%22%3A%22RANLP%202025%22%2C%22date%22%3A%222025-09%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%22%22%2C%22ISBN%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Furldefense.com%5C%2Fv3%5C%2F__https%3A%5C%2F%5C%2Fscholar.google.com%5C%2Fscholar_url%3Furl%3Dhttps%3A%2A%2AAacl-bg.org%2Aproceedings%2A2025%2ARANLP%2A25202025%2Apdf%2A2025.ranlp-1.7.pdf%26hl%3Den%26sa%3DX%26d%3D10294159676986260070%26ei%3DB0joaKPVHZ2qieoP8IbmuQU%26scisig%3DAAZF9b9cejTPsFaXQs7mHZOQ7ZN5%26oi%3Dscholaralrt%26hist%3DEEEsEVQAAAAJ%3A402328248577220524%3AAAZF9b8yqsrNs5DUnZheNhCdQQdp%26html%3D%26pos%3D0%26folt%3Dkw__%3BLy8vLy8lLy8%21%21DZ3fjg%215ghFLdOZTAyFeq9UESd1yTPa4blGhf6pTXFTPphU6DGQvGDjctFwclvNRJjuAiuQuJLtL7KVCYL6ZdNAeR2EIde2DJvcqAs%24%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T16%3A37%3A05Z%22%7D%7D%2C%7B%22key%22%3A%22286GCS4S%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Arora%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BArora%2C%20S.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Chain-of-Thought%20Reasoning%20in%20Streaming%20Full-Duplex%20End-to-End%20Spoken%20Dialogue%20Systems.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.02066%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.02066%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Chain-of-Thought%20Reasoning%20in%20Streaming%20Full-Duplex%20End-to-End%20Spoken%20Dialogue%20Systems%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Siddhant%22%2C%22lastName%22%3A%22Arora%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jinchuan%22%2C%22lastName%22%3A%22Tian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hayato%22%2C%22lastName%22%3A%22Futami%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiatong%22%2C%22lastName%22%3A%22Shi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yosuke%22%2C%22lastName%22%3A%22Kashiwagi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emiru%22%2C%22lastName%22%3A%22Tsunoo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shinji%22%2C%22lastName%22%3A%22Watanabe%22%7D%5D%2C%22abstractNote%22%3A%22Most%20end-to-end%20%28E2E%29%20spoken%20dialogue%20systems%20%28SDS%29%20rely%20on%20voice%20activity%20detection%20%28VAD%29%20for%20turn-taking%2C%20but%20VAD%20fails%20to%20distinguish%20between%20pauses%20and%20turn%20completions.%20Duplex%20SDS%20models%20address%20this%20by%20predicting%20output%20continuously%2C%20including%20silence%20tokens%2C%20thus%20removing%20the%20need%20for%20explicit%20VAD.%20However%2C%20they%20often%20have%20complex%20dual-channel%20architecture%20and%20lag%20behind%20cascaded%20models%20in%20semantic%20reasoning.%20To%20overcome%20these%20challenges%2C%20we%20propose%20SCoT%3A%20a%20Streaming%20Chain-of-Thought%20%28CoT%29%20framework%20for%20Duplex%20SDS%2C%20alternating%20between%20processing%20fixed-duration%20user%20input%20and%20generating%20responses%20in%20a%20blockwise%20manner.%20Using%20frame-level%20alignments%2C%20we%20create%20intermediate%20targets-aligned%20user%20transcripts%20and%20system%20responses%20for%20each%20block.%20Experiments%20show%20that%20our%20approach%20produces%20more%20coherent%20and%20interpretable%20responses%20than%20existing%20duplex%20methods%20while%20supporting%20lower-latency%20and%20overlapping%20interactions%20compared%20to%20turn-by-turn%20systems.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.02066%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.02066%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T16%3A15%3A34Z%22%7D%7D%2C%7B%22key%22%3A%22ND777N8X%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Yan%20and%20Snir%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BYan%2C%20J.%20%26amp%3B%20Snir%2C%20M.%20Contemplating%20a%26%23xA0%3BLightweight%20Communication%20Interface%20for%26%23xA0%3BAsynchronous%20Many-Task%20Systems.%20in%20%26lt%3Bi%26gt%3BAsynchronous%20Many-Task%20Systems%20and%20Applications%26lt%3B%5C%2Fi%26gt%3B%20%28eds%20Diehl%2C%20P.%2C%20Cao%2C%20Q.%2C%20Herault%2C%20T.%20%26amp%3B%20Bosilca%2C%20G.%29%20165%26%23x2013%3B173%20%28Springer%20Nature%20Switzerland%2C%20Cham%2C%202026%29.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2F978-3-031-97196-9_14%26%23039%3B%26gt%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2F978-3-031-97196-9_14%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Contemplating%20a%5Cu00a0Lightweight%20Communication%20Interface%20for%5Cu00a0Asynchronous%20Many-Task%20Systems%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiakun%22%2C%22lastName%22%3A%22Yan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marc%22%2C%22lastName%22%3A%22Snir%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Patrick%22%2C%22lastName%22%3A%22Diehl%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Qinglei%22%2C%22lastName%22%3A%22Cao%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Thomas%22%2C%22lastName%22%3A%22Herault%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22George%22%2C%22lastName%22%3A%22Bosilca%22%7D%5D%2C%22abstractNote%22%3A%22Asynchronous%20Many-Task%20Systems%20%28AMTs%29%20exhibit%20different%20communication%20patterns%20from%20traditional%20High-Performance%20Computing%20%28HPC%29%20applications%2C%20characterized%20by%20asynchrony%2C%20concurrency%2C%20and%20multithreading.%20Existing%20communication%20libraries%20usually%20do%20not%20support%20AMTs%5Cu2019%20communication%20requirements%20in%20the%20most%20direct%20and%20efficient%20ways.%20The%20Lightweight%20Communication%20Interface%20%28LCI%29%20is%20an%20experimental%20communication%20library%20aiming%20to%20push%20for%20efficient%20communication%20support%20for%20AMTs.%20This%20paper%20presents%20the%20design%20for%20a%20new%20LCI%20C%2B%2B%20interface%20and%20its%20rationale.%20With%20a%20new%20C%2B%2B%20objectized%20flexible%20functions%20idiom%2C%20the%20new%20interface%20aims%20for%20the%20following%20features%3A%20%28a%29%20a%20concise%20but%20expressive%20interface%20for%20all%20common%20point-to-point%20communication%20primitives%20and%20completion%20mechanisms%2C%20%28b%29%20a%20fine-grained%20resource%20mapping%20scheme%20for%20library%20interoperation%2C%20multithreaded%20performance%20isolation%2C%20and%20flexibility%20%28c%29%20a%20set%20of%20optional%20parameters%20and%20overridable%20classes%20for%20users%20to%20incrementally%20fine-tune%20the%20runtime%20behavior.%22%2C%22proceedingsTitle%22%3A%22Asynchronous%20Many-Task%20Systems%20and%20Applications%22%2C%22conferenceName%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1007%5C%2F978-3-031-97196-9_14%22%2C%22ISBN%22%3A%229783031971969%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T15%3A30%3A21Z%22%7D%7D%2C%7B%22key%22%3A%22JAJ75HAR%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Su%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BSu%2C%20B.-H.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Reasoning%20Beyond%20Majority%20Vote%3A%20An%20Explainable%20SpeechLM%20Framework%20for%20Speech%20Emotion%20Recognition.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.24187%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.24187%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Reasoning%20Beyond%20Majority%20Vote%3A%20An%20Explainable%20SpeechLM%20Framework%20for%20Speech%20Emotion%20Recognition%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bo-Hao%22%2C%22lastName%22%3A%22Su%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hui-Ying%22%2C%22lastName%22%3A%22Shih%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jinchuan%22%2C%22lastName%22%3A%22Tian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiatong%22%2C%22lastName%22%3A%22Shi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chi-Chun%22%2C%22lastName%22%3A%22Lee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Carlos%22%2C%22lastName%22%3A%22Busso%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shinji%22%2C%22lastName%22%3A%22Watanabe%22%7D%5D%2C%22abstractNote%22%3A%22Speech%20Emotion%20Recognition%20%28SER%29%20is%20typically%20trained%20and%20evaluated%20on%20majority-voted%20labels%2C%20which%20simplifies%20benchmarking%20but%20masks%20subjectivity%20and%20provides%20little%20transparency%20into%20why%20predictions%20are%20made.%20This%20neglects%20valid%20minority%20annotations%20and%20limits%20interpretability.%20We%20propose%20an%20explainable%20Speech%20Language%20Model%20%28SpeechLM%29%20framework%20that%20frames%20SER%20as%20a%20generative%20reasoning%20task.%20Given%20an%20utterance%2C%20the%20model%20first%20produces%20a%20transcript%2C%20then%20outputs%20both%20an%20emotion%20label%20and%20a%20concise%20natural-language%20rationale%20grounded%20in%20lexical%20and%20acoustic%20cues.%20Rationales%20are%20generated%20by%20a%20reasoning-capable%20teacher%20LLM%20and%20used%20as%20intermediate%20supervision%2C%20combined%20with%20majority%20labels%20during%20fine-tuning.%20Unlike%20prior%20work%20primarily%20focused%20on%20boosting%20classification%20accuracy%2C%20we%20aim%20to%20enhance%20explainability%20while%20preserving%20competitive%20performance.%20To%20this%20end%2C%20we%20complement%20majority-label%20metrics%20with%20annotator-aware%20scoring%20that%20credits%20matches%20with%20any%20annotator%20label.%20On%20MSP-Podcast%20v1.12%2C%20our%20model%20maintains%20improvements%20over%20zero-shot%20SpeechLM%20baselines%2C%20and%20produces%20rationales%20that%20human%20evaluators%20find%20plausible%20and%20well%20grounded.%20This%20demonstrates%20that%20incorporating%20rationale%20supervision%20offers%20a%20practical%20path%20toward%20interpretable%20SER%20without%20sacrificing%20predictive%20quality.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.24187%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.24187%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T15%3A09%3A35Z%22%7D%7D%2C%7B%22key%22%3A%22425YCH4Q%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Habib%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHabib%2C%20S.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Error%20quantification%20and%20comparison%20of%20binary%20neutron%20star%20gravitational%20waveforms%20from%20numerical%20relativity%20codes.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.23028%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.23028%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Error%20quantification%20and%20comparison%20of%20binary%20neutron%20star%20gravitational%20waveforms%20from%20numerical%20relativity%20codes%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sarah%22%2C%22lastName%22%3A%22Habib%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Elias%20R.%22%2C%22lastName%22%3A%22Most%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nils%22%2C%22lastName%22%3A%22Deppe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Francois%22%2C%22lastName%22%3A%22Foucart%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mark%20A.%22%2C%22lastName%22%3A%22Scheel%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Saul%22%2C%22lastName%22%3A%22Teukolsky%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%22%2C%22lastName%22%3A%22Boyle%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Matthew%22%2C%22lastName%22%3A%22Duez%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Larry%22%2C%22lastName%22%3A%22Kidder%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jordan%22%2C%22lastName%22%3A%22Moxon%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kyle%20C.%22%2C%22lastName%22%3A%22Nelli%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Harald%22%2C%22lastName%22%3A%22Pfeiffer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Throwe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nils%22%2C%22lastName%22%3A%22Vu%22%7D%5D%2C%22abstractNote%22%3A%22Future%20gravitational%20wave%20detections%20of%20merging%20binary%20neutron%20star%20systems%20have%20the%20possibility%20to%20tightly%20constrain%20the%20equation%20of%20state%20of%20dense%20nuclear%20matter.%20In%20order%20to%20extract%20such%20constraints%2C%20gravitational%20waveform%20models%20need%20to%20be%20calibrated%20to%20accurate%20numerical%20relativity%20simulations%20of%20the%20late%20inspiral%20and%20merger.%20In%20this%20work%2C%20we%20take%20an%20essential%20step%20toward%20classifying%20the%20error%20and%20potential%20systematics%20in%20current%20generation%20numerical%20relativity%20simulations%20of%20merging%20binary%20neutron%20stars.%20To%20this%20end%2C%20we%20perform%20a%20direct%20comparison%20of%20two%20codes%20%28FIL%2C%20SpEC%29%2C%20which%20differ%20in%20many%20aspects%2C%20including%20the%20numerical%20methods%20and%20discretizations%20used%20and%20equations%20solved.%20We%20find%20that%20despite%20these%20different%20approaches%2C%20the%20codes%20are%20--%20within%20current%20numerical%20resolution%20bounds%20--%20fully%20consistent%2C%20and%20broadly%20comparable%20in%20cost%20for%20a%20given%20accuracy%20level.%20Our%20results%20indicate%20that%20the%20error%20in%20the%20waveforms%20is%20primarily%20dominated%20by%20the%20hydrodynamic%20evolution%2C%20consistent%20with%20earlier%20findings%20in%20the%20literature.%20We%20also%20discuss%20current%20limitations%20and%20cost%20estimates%20for%20numerical%20relativity%20simulations%20to%20reach%20the%20accuracies%20required%20in%20the%20era%20of%20next-generation%20gravitational%20wave%20detectors.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.23028%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.23028%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T15%3A07%3A30Z%22%7D%7D%2C%7B%22key%22%3A%22WE9EMEZQ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zi%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZi%2C%20Y.%2C%20Wu%2C%20Z.%2C%20Boruch-Gruszecki%2C%20A.%2C%20Bell%2C%20J.%20%26amp%3B%20Guha%2C%20A.%20AgentPack%3A%20A%20Dataset%20of%20Code%20Changes%2C%20Co-Authored%20by%20Agents%20and%20Humans.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.21891%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.21891%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22AgentPack%3A%20A%20Dataset%20of%20Code%20Changes%2C%20Co-Authored%20by%20Agents%20and%20Humans%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yangtian%22%2C%22lastName%22%3A%22Zi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zixuan%22%2C%22lastName%22%3A%22Wu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aleksander%22%2C%22lastName%22%3A%22Boruch-Gruszecki%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jonathan%22%2C%22lastName%22%3A%22Bell%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Arjun%22%2C%22lastName%22%3A%22Guha%22%7D%5D%2C%22abstractNote%22%3A%22Fine-tuning%20large%20language%20models%20for%20code%20editing%20has%20typically%20relied%20on%20mining%20commits%20and%20pull%20requests.%20The%20working%20hypothesis%20has%20been%20that%20commit%20messages%20describe%20human%20intent%20in%20natural%20language%2C%20and%20patches%20to%20code%20describe%20the%20changes%20that%20implement%20that%20intent.%20However%2C%20much%20of%20the%20previously%20collected%20data%20is%20noisy%3A%20commit%20messages%20are%20terse%2C%20human-written%20commits%20commingle%20several%20unrelated%20edits%2C%20and%20many%20commits%20come%20from%20simple%2C%20rule-based%20bots.%5Cn%20The%20recent%20adoption%20of%20software%20engineering%20agents%20changes%20this%20landscape.%20Code%20changes%20co-authored%20by%20humans%20and%20agents%20tend%20to%20be%20more%20narrowly%20scoped%20and%20focused%20on%20clearer%20goals.%20Their%20commit%20messages%2C%20generated%20by%20LLMs%2C%20articulate%20intent%20and%20rationale%20in%20much%20greater%20detail.%20Moreover%2C%20when%20these%20changes%20land%20in%20public%20repositories%2C%20they%20are%20implicitly%20filtered%20by%20humans%3A%20maintainers%20discard%20low-quality%20commits%20to%20their%20projects.%5Cn%20We%20present%20AgentPack%2C%20a%20corpus%20of%201.3M%20code%20edits%20co-authored%20by%20Claude%20Code%2C%20OpenAI%20Codex%2C%20and%20Cursor%20Agent%20across%20public%20GitHub%20projects%20up%20to%20mid-August%202025.%20We%20describe%20the%20identification%20and%20curation%20pipeline%2C%20quantify%20adoption%20trends%20of%20these%20agents%2C%20and%20analyze%20the%20structural%20properties%20of%20the%20edits.%20Finally%2C%20we%20show%20that%20models%20fine-tuned%20on%20AgentPack%20can%20outperform%20models%20trained%20on%20prior%20human-only%20commit%20corpora%2C%20highlighting%20the%20potential%20of%20using%20public%20data%20from%20software%20engineering%20agents%20to%20train%20future%20code-editing%20models.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.21891%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.21891%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T15%3A05%3A01Z%22%7D%7D%2C%7B%22key%22%3A%22WM7IWST9%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Bhatnagar%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BBhatnagar%2C%20S.%2C%20Xu%2C%20A.%2C%20Tan%2C%20K.-H.%20%26amp%3B%20Ahuja%2C%20N.%20LUQ%3A%20Layerwise%20Ultra-Low%20Bit%20Quantization%20for%20Multimodal%20Large%20Language%20Models.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.23729%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.23729%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22LUQ%3A%20Layerwise%20Ultra-Low%20Bit%20Quantization%20for%20Multimodal%20Large%20Language%20Models%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shubhang%22%2C%22lastName%22%3A%22Bhatnagar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andy%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kar-Han%22%2C%22lastName%22%3A%22Tan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Narendra%22%2C%22lastName%22%3A%22Ahuja%22%7D%5D%2C%22abstractNote%22%3A%22Large%20Language%20Models%20%28LLMs%29%20with%20multimodal%20capabilities%20have%20revolutionized%20vision-language%20tasks%2C%20but%20their%20deployment%20often%20requires%20huge%20memory%20and%20computational%20resources.%20While%20post-training%20quantization%20%28PTQ%29%20has%20successfully%20compressed%20language%20models%20to%20as%20low%20as%201-bit%20precision%20without%20significant%20performance%20loss%2C%20its%20effectiveness%20for%20multimodal%20LLMs%20%28MLLMs%29%20remains%20relatively%20unexplored.%20In%20this%20paper%2C%20we%20present%20the%20first%20study%20on%20ultra-low%20bit%20%28%26lt%3B4-bit%29%20quantization%20for%20multimodal%20LLMs.%20Our%20analysis%20reveals%20that%20multimodal%20tokens%20and%20intermediate%20layer%20activations%20produced%20by%20them%20exhibit%20significantly%20higher%20statistical%20variance%20and%20entropy%20compared%20to%20text%20tokens%2C%20making%20them%20less%20tolerant%20to%20ultra-low%20bit%20quantization.%20However%2C%20the%20activation%20distributions%20of%20multimodal%20tokens%20varies%20significantly%20over%20different%20layers%2C%20with%20some%20layers%20having%20lower%20entropy%20activation%20distributions.%20We%20empirically%20show%20that%20such%20layers%20in%20these%20models%20can%20better%20tolerate%20ultra-low%20bit%20quantization.%20Building%20on%20these%20insights%2C%20we%20propose%20a%20novel%20strategy%20for%20MLLM%20quantization%2C%20LUQ%3A%20Layerwise%20Ultra-Low%20Bit%20Quantization%2C%20which%20selectively%20applies%20ultra-low%20bit%20quantization%20to%20layers%20that%20are%20more%20resilient%20to%20it.%20Additionally%2C%20we%20also%20show%20that%20using%20a%20mix%20of%20multimodal%20tokens%20%28image%20and%20text%29%20for%20PTQ%20boosts%20VQA%20performance%20in%20the%20ultra-low%20bit%20regime.%20We%20evaluate%20our%20method%20on%20LLaVA-1.5%20and%20Qwen-2.5-VL%20across%209%20popular%20VQA%20benchmarks.%20The%20resulting%20LUQ%20models%20use%2040%25%20and%2031%25%20less%20memory%20than%20their%204-bit%20counterparts%2C%20respectively%2C%20while%20exhibiting%20a%20performance%20degradation%20of%20less%20than%2010%25%20on%20the%20MME%20benchmark.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.23729%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.23729%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T15%3A02%3A43Z%22%7D%7D%2C%7B%22key%22%3A%22YLZBA5ME%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Srivastava%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BSrivastava%2C%20G.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20BeyondBench%3A%20Benchmark-Free%20Evaluation%20of%20Reasoning%20in%20Language%20Models.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.24210%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.24210%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22BeyondBench%3A%20Benchmark-Free%20Evaluation%20of%20Reasoning%20in%20Language%20Models%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Gaurav%22%2C%22lastName%22%3A%22Srivastava%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aafiya%22%2C%22lastName%22%3A%22Hussain%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhenyu%22%2C%22lastName%22%3A%22Bi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Swastik%22%2C%22lastName%22%3A%22Roy%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Priya%22%2C%22lastName%22%3A%22Pitre%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Meng%22%2C%22lastName%22%3A%22Lu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Morteza%22%2C%22lastName%22%3A%22Ziyadi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xuan%22%2C%22lastName%22%3A%22Wang%22%7D%5D%2C%22abstractNote%22%3A%22Evaluating%20language%20models%20fairly%20is%20becoming%20harder%20as%20static%20benchmarks%20available%20on%20the%20internet%20risk%20contamination%20by%20training%20data.%20This%20makes%20it%20unclear%20whether%20models%20are%20truly%20reasoning%20or%20just%20recalling%20answers.%20In%20this%20paper%2C%20we%20introduce%20BeyondBench%2C%20an%20evaluation%20framework%20that%20avoids%20this%20problem%20by%20using%20algorithmic%20problem%20generation.%20Unlike%20traditional%20benchmarks%20that%20risk%20contamination%20from%20internet-scale%20training%20data%2C%20BeyondBench%20creates%20mathematically%20grounded%20problems%20on%20the%20fly%2C%20ensuring%20each%20test%20remains%20fresh%20and%20uncontaminated.%20Our%20framework%20covers%2044%20algorithmic%20tasks%20with%20a%20total%20of%20117%20variations%2C%20grouped%20into%20three%20difficulty%20levels%3A%20the%20Easy%20Suite%20%2829%20tasks%29%20for%20basic%20arithmetic%20and%20statistics%2C%20the%20Medium%20Suite%20%285%20tasks%2C%2049%20variations%29%20for%20sequence%20patterns%20and%20reasoning%2C%20and%20the%20Hard%20Suite%20%2810%20tasks%2C%2068%20variations%29%20tackling%20NP-complete%20and%20constraint%20satisfaction%20problems.%20Each%20task%20generates%20problems%20from%20a%20combinatorial%20space%20larger%20than%2010%5E15%20unique%20instances%2C%20with%20solutions%20verified%20deterministically%20by%20mathematical%20proofs.%20We%20evaluated%20101%20language%20models%2C%20including%2085%20open-source%20and%2016%20closed-source%20models%2C%20spanning%20sizes%20from%200.5B%20to%20141B%20parameters%20and%20multiple%20quantization%20schemes.%20Our%20results%20show%20consistent%20reasoning%20deficiencies%20across%20model%20families%2C%20with%20performance%20degrading%20sharply%20as%20problem%20complexity%20increases%20from%20polynomial%20to%20exponential.%20In%20our%20Hard%20Suite%20evaluations%2C%20models%20such%20as%20Gemini-2.5-pro%2C%20Llama-3.3-70B%2C%20and%20Qwen2.5-72B%20achieved%20average%20accuracies%20of%2056.38%25%2C%2026.91%25%2C%20and%2033.60%25%2C%20respectively.%20Moreover%2C%20we%20observe%20that%20performance%20drops%20drastically%20without%20tool%20usage%2C%20with%20GPT-5%2C%20GPT-5-mini%2C%20and%20GPT-5-nano%20showing%20a%20decline%20of%2016.81%25%2C%2028.05%25%2C%20and%2047.59%25%20accuracy%20on%20the%20hard%20suite.%20Our%20leaderboard%20is%20publicly%20available%20at%20https%3A%5C%2F%5C%2Fctrl-gaurav.github.io%5C%2FBeyondBench%5C%2F%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.24210%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.24210%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T14%3A58%3A35Z%22%7D%7D%2C%7B%22key%22%3A%22PF4AYG5G%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Marx%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMarx%2C%20E.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Likelihood-free%20inference%20for%20gravitational-wave%20data%20analysis%20and%20public%20alerts.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.22561%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.22561%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Likelihood-free%20inference%20for%20gravitational-wave%20data%20analysis%20and%20public%20alerts%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ethan%22%2C%22lastName%22%3A%22Marx%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Deep%22%2C%22lastName%22%3A%22Chatterjee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Malina%22%2C%22lastName%22%3A%22Desai%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ravi%22%2C%22lastName%22%3A%22Kumar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Benoit%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Argyro%22%2C%22lastName%22%3A%22Sasli%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Leo%22%2C%22lastName%22%3A%22Singer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%20W.%22%2C%22lastName%22%3A%22Coughlin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Philip%22%2C%22lastName%22%3A%22Harris%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Erik%22%2C%22lastName%22%3A%22Katsavounidis%22%7D%5D%2C%22abstractNote%22%3A%22Rapid%20and%20reliable%20detection%20and%20dissemination%20of%20source%20parameter%20estimation%20data%20products%20from%20gravitational-wave%20events%2C%20especially%20sky%20localization%2C%20is%20critical%20for%20maximizing%20the%20potential%20of%20multi-messenger%20astronomy.%20Machine%20learning%20based%20detection%20and%20parameter%20estimation%20algorithms%20are%20emerging%20as%20production%20ready%20alternatives%20to%20traditional%20approaches.%20Here%2C%20we%20report%20validation%20studies%20of%20AMPLFI%2C%20a%20likelihood-free%20inference%20solution%20to%20low-latency%20parameter%20estimation%20of%20binary%20black%20holes.%20We%20use%20simulated%20signals%20added%20into%20data%20from%20the%20LIGO-Virgo-KAGRA%26%23039%3Bs%20%28LVK%26%23039%3Bs%29%20third%20observing%20run%20%28O3%29%20to%20compare%20sky%20localization%20performance%20with%20BAYESTAR%2C%20the%20algorithm%20currently%20in%20production%20for%20rapid%20sky%20localization%20of%20candidates%20from%20matched-filter%20pipelines.%20We%20demonstrate%20sky%20localization%20performance%2C%20measured%20by%20searched%20area%20and%20volume%2C%20to%20be%20equivalent%20with%20BAYESTAR.%20We%20show%20accurate%20reconstruction%20of%20source%20parameters%20with%20uncertainties%20for%20use%20distributing%20low-latency%20coarse-grained%20chirp%20mass%20information.%20In%20addition%2C%20we%20analyze%20several%20candidate%20events%20reported%20by%20the%20LVK%20in%20the%20third%20gravitational-wave%20transient%20catalog%20%28GWTC-3%29%20and%20show%20consistency%20with%20the%20LVK%26%23039%3Bs%20analysis.%20Altogether%2C%20we%20demonstrate%20AMPLFI%26%23039%3Bs%20ability%20to%20produce%20data%20products%20for%20low-latency%20public%20alerts.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.22561%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.22561%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T14%3A56%3A37Z%22%7D%7D%2C%7B%22key%22%3A%22YEVCLFUK%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Liu%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-26%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLiu%2C%20Z.%2C%20Gong%2C%20A.%20%26amp%3B%20Clark%2C%20B.%20K.%20Decoding%20quantum%20low%20density%20parity%20check%20codes%20with%20diffusion.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2509.22347%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2509.22347%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Decoding%20quantum%20low%20density%20parity%20check%20codes%20with%20diffusion%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zejun%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Anqi%22%2C%22lastName%22%3A%22Gong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bryan%20K.%22%2C%22lastName%22%3A%22Clark%22%7D%5D%2C%22abstractNote%22%3A%22An%20efficient%20decoder%20is%20essential%20for%20quantum%20error%20correction%2C%20and%20data-driven%20neural%20decoders%20have%20emerged%20as%20promising%2C%20flexible%20solutions.%20Here%2C%20we%20introduce%20a%20diffusion%20model%20framework%20to%20infer%20logical%20errors%20from%20syndrome%20measurements%20in%20quantum%20low-density%20parity-check%20codes.%20Using%20the%20bivariate%20bicycle%20code%20with%20realistic%20circuit-level%20noise%2C%20we%20show%20that%20masked%20diffusion%20decoders%20are%20more%20accurate%2C%20often%20faster%20on%20average%2C%20and%20always%20faster%20in%20the%20worst%20case%20than%20other%20state-of-the-art%20decoders%2C%20including%20belief%20propagation%20with%20ordered%20statistics%20decoding%20%28BP-OSD%29%20and%20autoregressive%20neural%20decoders.%20We%20show%20that%20by%20using%20fewer%20diffusion%20steps%20during%20inference%20one%20can%20gain%20significant%20speed%20at%20minimal%20cost%20in%20accuracy.%20By%20examining%20the%20factored%20attention%20from%20our%20trained%20neural%20network%20we%20find%20that%2C%20despite%20being%20trained%20solely%20on%20paired%20samples%20of%20syndrome-logical%20errors%2C%20this%20diffusion%20decoder%20learns%20the%20structure%20of%20the%20quantum%20codes.%20We%20also%20compare%20both%20masked%20and%20continuous%20diffusion%20decoders%20on%20code-capacity%20noise%20models%2C%20finding%20that%20masked%20diffusion%20decoders%20scale%20better%20than%20continuous%20diffusion%20decoders.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2509.22347%22%2C%22date%22%3A%222025-09-26%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2509.22347%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.22347%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T14%3A46%3A43Z%22%7D%7D%2C%7B%22key%22%3A%22F7K4LR28%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Halevi%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-19%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHalevi%2C%20G.%2C%20Shankar%2C%20S.%2C%20M%26%23xF6%3Bsta%2C%20P.%2C%20Haas%2C%20R.%20%26amp%3B%20Schnetter%2C%20E.%20A%20Black%20Hole%20is%20Born%3A%203D%20GRMHD%20Simulation%20of%20Black%20Hole%20Formation%20from%20Core-Collapse.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2506.20837%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2506.20837%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22A%20Black%20Hole%20is%20Born%3A%203D%20GRMHD%20Simulation%20of%20Black%20Hole%20Formation%20from%20Core-Collapse%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Goni%22%2C%22lastName%22%3A%22Halevi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Swapnil%22%2C%22lastName%22%3A%22Shankar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Philipp%22%2C%22lastName%22%3A%22M%5Cu00f6sta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Roland%22%2C%22lastName%22%3A%22Haas%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Erik%22%2C%22lastName%22%3A%22Schnetter%22%7D%5D%2C%22abstractNote%22%3A%22We%20present%20the%20first%20three-dimensional%2C%20fully%20general-relativistic%20magnetohydrodynamic%20%283D%20GRMHD%29%20simulation%20of%20a%20black%20hole%20%28BH%29%20formed%20from%20the%20collapsed%20core%20of%20a%20massive%20star.%20The%20ability%20to%20self-consistently%20capture%20the%20birth%20of%20a%20compact%20remnant%20in%203D%20is%20crucial%20for%20modeling%20natal%20BH%20properties%20%28including%20masses%2C%20spins%2C%20and%20kicks%29%2C%20which%20are%20of%20particular%20interest%20in%20the%20era%20of%20gravitational%20wave%20astronomy.%20However%2C%20such%20simulations%20have%20remained%20elusive%20due%20to%20extreme%20computational%20challenges%20and%20demands.%20We%20employ%20the%20GPU-accelerated%20dynamical-spacetime%20GRMHD%20code%20GRaM-X%20to%20follow%20the%20collapse%2C%20core-bounce%2C%20shock%20propagation%2C%20and%20eventual%20BH%20formation%20of%20a%20massive%20stellar%20progenitor%20in%20full%203D.%20We%20initialize%20our%20simulation%20by%20mapping%20a%20one-dimensional%20%281D%29%20model%20of%20a%20star%20with%20a%20zero-age-main-sequence%20mass%20of%20%2445%20M_%5C%5Codot%24%20to%203D.%20We%20use%20moderate%20rotation%20%28consistent%20with%20expectations%20from%20stellar%20evolution%20modeling%29%20and%20a%20relatively%20weak%20dipolar%20magnetic%20field.%20The%20collapsing%20core%20drives%20a%20shock%20that%20reaches%20a%20maximum%20radius%20of%20roughly%20170%20km%20before%20stalling%20and%20does%20not%20lead%20to%20a%20successful%20explosion.%20The%20proto-neutron%20star%20accretes%20matter%20before%20collapsing%20to%20form%20a%20BH%20%24t_%7BBH%7D%20%5C%5Capprox%20325%24%20ms%20after%20core-bounce.%20The%20time%20of%20BH%20formation%20and%20initial%20BH%20mass%20are%20remarkably%20similar%20to%20those%20obtained%20with%20GR1D%2C%20a%201D%20general-relativistic%20neutrino-hydrodynamics%20code%2C%20to%20which%20we%20compare%20our%20results.%20We%20track%20the%20horizon%20of%20the%20newborn%20BH%20after%20formation%20and%20calculate%20a%20steady%20kick%20velocity%20of%20%24v_%7Bkick%7D%20%5C%5Capprox%2072%24%20km%5C%2Fs%20and%20a%20mass%20of%20%24M_%7BBH%7D%20%5C%5Capprox%202.62%20M_%5C%5Codot%24%2C%20which%20is%20still%20rising%20at%20the%20end%20of%20the%20simulation.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2506.20837%22%2C%22date%22%3A%222025-09-19%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2506.20837%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2506.20837%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T14%3A41%3A16Z%22%7D%7D%2C%7B%22key%22%3A%228UDKVEU3%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lewis%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-26%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLewis%2C%20A.%2C%20Himberg%2C%20B.%2C%20Torres-S%26%23xE1%3Bnchez%2C%20A.%20%26amp%3B%20Vanegas%2C%20J.%20Microscopic%20elasticity%20from%20MD%20part%20I%3A%20Bulk%20solid%20and%20fluid%20systems.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv-2025-j17x4%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv-2025-j17x4%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Microscopic%20elasticity%20from%20MD%20part%20I%3A%20Bulk%20solid%20and%20fluid%20systems%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andrew%22%2C%22lastName%22%3A%22Lewis%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Benjamin%22%2C%22lastName%22%3A%22Himberg%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alejandro%22%2C%22lastName%22%3A%22Torres-S%5Cu00e1nchez%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Juan%22%2C%22lastName%22%3A%22Vanegas%22%7D%5D%2C%22abstractNote%22%3A%22Computational%20modeling%20such%20as%20molecular%20dynamics%20%28MD%29%20and%20Monte%20Carlo%20simulations%20can%20be%20used%20to%20estimate%20the%20elastic%20properties%20of%20materials%20through%20various%20stress%20and%20strain%20relationships.%20Here%2C%20we%20demonstrate%20the%20effectiveness%20of%20the%20stress-stress%20fluctuation%20%28SSF%29%20method%20to%20estimate%20the%20elastic%20properties%20of%20simple%20van%20der%20Waals%20and%20molecular%20materials.%20Starting%20with%20argon%20in%20the%20solid%2C%20liquid%2C%20and%20gas%20phases%2C%20we%20show%20that%20the%20SSF%20method%20gives%20elastic%20coefficients%20and%20moduli%20in%20excellent%20agreement%20with%20values%20obtained%20with%20the%20explicit%20deformation%20and%20volume%20fluctuation%20methods.%20Comparison%20of%20the%20elastic%20coefficients%20and%20bulk%20modulus%20for%20solid%20argon%20with%20previous%20computational%20studies%20and%20experimental%20data%20provide%20further%20validation%20of%20our%20numerical%20implementation.%20Beyond%20argon%2C%20we%20show%20that%20the%20elastic%20properties%20of%20molecular%20fluids%20simulated%20with%20the%20CG%20MARTINI%20force-field%2C%20which%20include%20multi-body%20interactions%20such%20as%20angle%20potentials%2C%20are%20also%20accurately%20captured%20by%20the%20SSF%20method.%20Moreover%2C%20the%20impulsive%20correction%20for%20truncated%20potentials%20is%20essential%20to%20obtain%20accurate%20values%20for%20these%20fluids%20and%20vanishing%20shear%20moduli.%20Our%20results%20highlight%20the%20broad%20applicability%20of%20the%20SSF%20method%2C%20which%20does%20not%20require%20any%20material%20deformation%2C%20across%20a%20broad%20range%20of%20systems%20and%20lays%20the%20foundation%20for%20its%20use%20to%20characterize%20the%20elastic%20properties%20of%20complex%20molecular%20systems.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025-09-26%22%2C%22DOI%22%3A%2210.26434%5C%2Fchemrxiv-2025-j17x4%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fchemrxiv.org%5C%2Fengage%5C%2Fchemrxiv%5C%2Farticle-details%5C%2F68d31c59f416303770093e16%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-08T14%3A33%3A50Z%22%7D%7D%2C%7B%22key%22%3A%22PNABIL4I%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zhao%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZhao%2C%20Y.%2C%20Wang%2C%20Z.%20%26amp%3B%20Zhang%2C%20M.%20PuzzleMoE%3A%20Efficient%20Compression%20of%20Large%20Mixture-of-Experts%20Models%20via%20Sparse%20Expert%20Merging%20and%20Bit-packed%20inference.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.04805%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.04805%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22PuzzleMoE%3A%20Efficient%20Compression%20of%20Large%20Mixture-of-Experts%20Models%20via%20Sparse%20Expert%20Merging%20and%20Bit-packed%20inference%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yushu%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zheng%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minjia%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22Mixture-of-Experts%20%28MoE%29%20models%20have%20shown%20strong%20potential%20in%20scaling%20language%20models%20efficiently%20by%20activating%20only%20a%20small%20subset%20of%20experts%20per%20input.%20However%2C%20their%20widespread%20deployment%20remains%20limited%20due%20to%20the%20high%20memory%20overhead%20associated%20with%20storing%20all%20expert%20parameters%2C%20particularly%20as%20the%20number%20of%20experts%20increases.%20To%20address%20this%20challenge%2C%20prior%20works%20have%20explored%20expert%20dropping%20and%20merging%20strategies%2C%20yet%20they%20often%20suffer%20from%20performance%20drop%20at%20high%20compression%20ratios.%20In%20this%20paper%2C%20we%20introduce%20PuzzleMoE%2C%20a%20training-free%20MoE%20compression%20method%20that%20achieves%20both%20high%20accuracy%20and%20efficient%20inference%20through%20two%20key%20innovations%3A%20First%2C%20PuzzleMoE%20performs%20sparse%20expert%20merging%20by%20identifying%20element-wise%20weight%20redundancy%20and%20specialization.%20It%20uses%20a%20dual-mask%20to%20capture%20both%20shared%20and%20expert-specific%20parameters.%20Second%2C%20to%20avoid%20the%20overhead%20of%20storing%20binary%20masks%20and%20signs%2C%20PuzzleMoE%20introduces%20a%20bit-packed%20encoding%20scheme%20that%20reuses%20underutilized%20exponent%20bits%2C%20enabling%20efficient%20MoE%20inference%20on%20GPUs.%20Extensive%20experiments%20demonstrate%20that%20PuzzleMoE%20can%20compress%20MoE%20models%20by%20up%20to%2050%25%20while%20maintaining%20accuracy%20across%20various%20tasks.%20Specifically%2C%20it%20outperforms%20prior%20MoE%20compression%20methods%20by%20up%20to%2016.7%25%20on%20MMLU%20at%2050%25%20compression%20ratio%2C%20and%20achieves%20up%20to%201.28%5C%5Ctimes%20inference%20speedup.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2511.04805%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2511.04805%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T23%3A17%3A48Z%22%7D%7D%2C%7B%22key%22%3A%22PFH3JFS7%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Loehr%20and%20Clark%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLoehr%2C%20K.%20%26amp%3B%20Clark%2C%20B.%20K.%20Enhancing%20Neural%20Network%20Backflow.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.26906%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.26906%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Enhancing%20Neural%20Network%20Backflow%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kieran%22%2C%22lastName%22%3A%22Loehr%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bryan%20K.%22%2C%22lastName%22%3A%22Clark%22%7D%5D%2C%22abstractNote%22%3A%22Accurately%20describing%20the%20ground%20state%20of%20strongly%20correlated%20systems%20is%20essential%20for%20understanding%20their%20emergent%20properties.%20Neural%20Network%20Backflow%20%28NNBF%29%20is%20a%20powerful%20variational%20ansatz%20that%20enhances%20mean-field%20wave%20functions%20by%20introducing%20configuration-dependent%20modifications%20to%20single-particle%20orbitals.%20Although%20NNBF%20is%20theoretically%20universal%20in%20the%20limit%20of%20large%20networks%2C%20we%20find%20that%20practical%20gains%20saturate%20with%20increasing%20network%20size.%20Instead%2C%20significant%20improvements%20can%20be%20achieved%20by%20using%20a%20multi-determinant%20ansatz.%20We%20explore%20efficient%20ways%20to%20generate%20these%20multi-determinant%20expansions%20without%20increasing%20the%20number%20of%20variational%20parameters.%20In%20particular%2C%20we%20study%20single-step%20Lanczos%20and%20symmetry%20projection%20techniques%2C%20benchmarking%20their%20performance%20against%20diffusion%20Monte%20Carlo%20and%20NNBF%20applied%20to%20alternative%20mean%20fields.%20Benchmarking%20on%20a%20doped%20periodic%20square%20Hubbard%20model%20near%20optimal%20doping%2C%20we%20find%20that%20a%20Lanczos%20step%2C%20diffusion%20Monte%20Carlo%2C%20and%20projection%20onto%20a%20symmetry%20sector%20all%20give%20similar%20improvements%20achieving%20state-of-the-art%20energies%20at%20minimal%20cost.%20By%20further%20optimizing%20the%20projected%20symmetrized%20states%20directly%2C%20we%20gain%20significantly%20in%20energy.%20Using%20this%20technique%20we%20report%20the%20lowest%20variational%20energies%20for%20this%20Hamiltonian%20on%20%244%5C%5Ctimes%2016%24%20and%20%244%20%5C%5Ctimes%208%24%20lattices%20as%20well%20as%20accurate%20variance%20extrapolated%20energies.%20We%20also%20show%20the%20evolution%20of%20spin%2C%20charge%2C%20and%20pair%20correlation%20functions%20as%20the%20quality%20of%20the%20variational%20ansatz%20improves.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.26906%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.26906%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T22%3A31%3A19Z%22%7D%7D%2C%7B%22key%22%3A%22SHLYZEQA%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Cheng%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCheng%2C%20C.-H.%2C%20Ficarra%2C%20G.%20%26amp%3B%20Witek%2C%20H.%20Dephasing%20in%20binary%20black%20hole%20mergers%20surrounded%20by%20scalar%20wave%20dark%20matter%20clouds.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.20037%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.20037%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Dephasing%20in%20binary%20black%20hole%20mergers%20surrounded%20by%20scalar%20wave%20dark%20matter%20clouds%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cheng-Hsin%22%2C%22lastName%22%3A%22Cheng%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Giuseppe%22%2C%22lastName%22%3A%22Ficarra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Helvi%22%2C%22lastName%22%3A%22Witek%22%7D%5D%2C%22abstractNote%22%3A%22Scalar%20fields%20of%20masses%20between%20%2410%5E%7B-21%7D%5C%5Crm%7BeV%7D%5C%2Fc%5E2%24%20and%20%2410%5E%7B-11%7D%20%5C%5Crm%7BeV%7D%5C%2Fc%5E2%24%20can%20exhibit%20enhanced%20gravitational%20interactions%20with%20black%20holes%2C%20and%20form%20scalar%20clouds%20around%20them.%20Such%20a%20cloud%20modifies%20the%20dynamics%20of%20a%20coalescing%20black-hole%20binary%2C%20and%20the%20resulting%20gravitational%20waves%20may%20provide%20a%20new%20channel%20to%20detect%20light%20scalar%20fields%2C%20such%20as%20axion-like%20particles%20or%20wave-like%20dark%20matter%20candidates.%20In%20this%20work%20we%20simulate%20a%20series%20of%20black-hole%20mergers%20with%20mass%20ratios%20%24q%3D1%24%20and%20%24q%3D1%5C%2F2%24%2C%20immersed%20in%20an%20scalar%20field%20overdensity%20with%20masses%20in%20the%20range%20%24M%5Cu03bc_%7B%5C%5Crm%7BS%7D%7D%20%5C%5Cin%5B0%2C1.0%5D%24.%20To%20do%20so%2C%20we%20implemented%20a%20constraint-satisfying%20initial%20data%20solver%20based%20on%20the%20puncture%20method%2C%20we%20improved%20the%20accuracy%20of%20our%20open-source%20software%20Canuda%20to%20eighth%20order%20finite%20differences%2C%20and%20we%20reduced%20the%20initial%20orbital%20eccentricity.%20We%20investigate%20the%20impact%20of%20the%20scalar%20mass%20on%20the%20gravitational%20and%20scalar%20radiation.%20We%20find%20that%20binaries%20can%20undergo%20a%20delayed%20or%20an%20accelerated%20merger%20with%20respect%20to%20the%20vacuum.%20Our%20study%20highlights%20the%20challenge%20and%20importance%20of%20accurately%20modeling%20black-hole%20binaries%20in%20dark%20matter%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.20037%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.20037%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T22%3A22%3A50Z%22%7D%7D%2C%7B%22key%22%3A%22Y68VDVBH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zhang%20et%20al.%22%2C%22parsedDate%22%3A%222025-10-29%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZhang%2C%20Z.%20A.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20One%20Token%20per%20Highly%20Selective%20Frame%3A%20Towards%20Extreme%20Compression%20for%20Long%20Video%20Understanding.%20in%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22One%20Token%20per%20Highly%20Selective%20Frame%3A%20Towards%20Extreme%20Compression%20for%20Long%20Video%20Understanding%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zheyu%20Aqa%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ziqi%22%2C%22lastName%22%3A%22Pang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shixing%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiang%22%2C%22lastName%22%3A%22Hao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Vimal%22%2C%22lastName%22%3A%22Bhat%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yu-Xiong%22%2C%22lastName%22%3A%22Wang%22%7D%5D%2C%22abstractNote%22%3A%22Long%20video%20understanding%20is%20inherently%20challenging%20for%20vision-language%20models%20%28VLMs%29%20because%20of%20the%20extensive%20number%20of%20frames.%20With%20each%20video%20frame%20typically%20expanding%20into%20tens%20or%20hundreds%20of%20tokens%2C%20the%20limited%20context%20length%20of%20large%20language%20models%20%28LLMs%29%20forces%20the%20VLMs%20to%20perceive%20the%20frames%20sparsely%20and%20lose%20temporal%20information.%20To%20address%20this%2C%20we%20explore%20extreme%20video%20token%20compression%20towards%20%2Aone%20token%20per%20frame%2A%20at%20the%20final%20LLM%20layer.%20Our%20key%20insight%20is%20that%20heuristic-based%20compression%2C%20widely%20adopted%20by%20previous%20methods%2C%20is%20prone%20to%20information%20loss%2C%20and%20this%20necessitates%20supervising%20LLM%20layers%20into%20%2Alearnable%2A%20and%20%2Aprogressive%2A%20modules%20for%20%2Atoken-level%20compression%2A%20%28LP-Comp%29.%20Such%20compression%20enables%20our%20VLM%20to%20digest%202x-4x%20more%20frames%20with%20improved%20performance.%20To%20further%20increase%20the%20token%20efficiency%2C%20we%20investigate%20%5C%5Cemph%7Bframe-level%20compression%7D%2C%20which%20selects%20the%20frames%20most%20relevant%20to%20the%20queries%20via%20the%20internal%20attention%20scores%20of%20the%20LLM%20layers%2C%20named%20%2Aquestion-conditioned%20compression%2A%20%28QC-Comp%29.%20As%20a%20notable%20distinction%20from%20previous%20studies%2C%20we%20mitigate%20the%20position%20bias%20of%20LLM%20attention%20in%20long%20contexts%2C%20%2Ai.e.%2A%2C%20the%20over-concentration%20on%20the%20beginning%20and%20end%20of%20a%20sequence%2C%20by%20splitting%20long%20videos%20into%20short%20segments%20and%20employing%20local%20attention.%20Collectively%2C%20our%20combined%20%2Atoken-level%2A%20and%20%2Aframe-level%2A%20leads%20to%20an%20e%2A%2Ax%2A%2Atreme%20compression%20model%20for%20long%20video%20understanding%2C%20named%20%2A%2AXComp%2A%2A%2C%20achieving%20a%20significantly%20larger%20compression%20ratio%20and%20enabling%20denser%20frame%20sampling.%20Our%20XComp%20is%20finetuned%20from%20VideoChat-Flash%20with%20a%20data-efficient%20%2Asupervised%20compression%20tuning%2A%20stage%20that%20only%20requires%202.5%5C%5C%25%20of%20the%20supervised%20fine-tuning%20data%2C%20yet%20boosts%20the%20accuracy%20from%2042.9%5C%5C%25%20to%2046.2%5C%5C%25%20on%20LVBench%20and%20enhances%20multiple%20other%20long%20video%20benchmarks.%22%2C%22proceedingsTitle%22%3A%22%22%2C%22conferenceName%22%3A%22The%20Thirty-ninth%20Annual%20Conference%20on%20Neural%20Information%20Processing%20Systems%22%2C%22date%22%3A%222025%5C%2F10%5C%2F29%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%22%22%2C%22ISBN%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fopenreview.net%5C%2Fforum%3Fid%3DbythzT0b81%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T22%3A13%3A24Z%22%7D%7D%2C%7B%22key%22%3A%22TJJPWREH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Cui%20et%20al.%22%2C%22parsedDate%22%3A%222025-11-16%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCui%2C%20S.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Story%20of%20Two%20GPUs%3A%20Characterizing%20the%20Resilience%20of%20Hopper%20H100%20and%20Ampere%20A100%20GPUs.%20in%20%26lt%3Bi%26gt%3BProceedings%20of%20the%20International%20Conference%20for%20High%20Performance%20Computing%2C%20Networking%2C%20Storage%20and%20Analysis%26lt%3B%5C%2Fi%26gt%3B%201145%26%23x2013%3B1164%20%28ACM%2C%20St.%20Louis%20MO%20USA%2C%202025%29.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1145%5C%2F3712285.3759821%26%23039%3B%26gt%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1145%5C%2F3712285.3759821%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Story%20of%20Two%20GPUs%3A%20Characterizing%20the%20Resilience%20of%20Hopper%20H100%20and%20Ampere%20A100%20GPUs%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shengkun%22%2C%22lastName%22%3A%22Cui%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Archit%22%2C%22lastName%22%3A%22Patke%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hung%22%2C%22lastName%22%3A%22Nguyen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aditya%22%2C%22lastName%22%3A%22Ranjan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ziheng%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Phuong%22%2C%22lastName%22%3A%22Cao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Gregory%22%2C%22lastName%22%3A%22Bauer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Brett%22%2C%22lastName%22%3A%22Bode%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Catello%20Di%22%2C%22lastName%22%3A%22Martino%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Saurabh%22%2C%22lastName%22%3A%22Jha%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chandra%22%2C%22lastName%22%3A%22Narayanaswami%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daby%22%2C%22lastName%22%3A%22Sow%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zbigniew%20T.%22%2C%22lastName%22%3A%22Kalbarczyk%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ravishankar%20K.%22%2C%22lastName%22%3A%22Iyer%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%22Proceedings%20of%20the%20International%20Conference%20for%20High%20Performance%20Computing%2C%20Networking%2C%20Storage%20and%20Analysis%22%2C%22conferenceName%22%3A%22SC%20%2725%3A%20The%20International%20Conference%20for%20High%20Performance%20Computing%2C%20Networking%2C%20Storage%20and%20Analysis%22%2C%22date%22%3A%222025-11-16%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1145%5C%2F3712285.3759821%22%2C%22ISBN%22%3A%229798400714665%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fdl.acm.org%5C%2Fdoi%5C%2F10.1145%5C%2F3712285.3759821%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T21%3A41%3A47Z%22%7D%7D%2C%7B%22key%22%3A%2247ALBZFP%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCross-Domain%20Long-Term%20Forecasting%3A%20Radiation%20Dose%20from%20Sparse%20Neutron%20Sensor%20via%20Spatio-Temporal%20Operator%20Network.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fhtml%5C%2F2510.18041v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fhtml%5C%2F2510.18041v1%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22Cross-Domain%20Long-Term%20Forecasting%3A%20Radiation%20Dose%20from%20Sparse%20Neutron%20Sensor%20via%20Spatio-Temporal%20Operator%20Network%22%2C%22creators%22%3A%5B%5D%2C%22abstractNote%22%3A%22%22%2C%22date%22%3A%22%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fhtml%5C%2F2510.18041v1%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T21%3A30%3A08Z%22%7D%7D%2C%7B%22key%22%3A%226YS366QL%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Chen%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BChen%2C%20H.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20ERA%3A%20Transforming%20VLMs%20into%20Embodied%20Agents%20via%20Embodied%20Prior%20Learning%20and%20Online%20Reinforcement%20Learning.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.12693%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.12693%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22ERA%3A%20Transforming%20VLMs%20into%20Embodied%20Agents%20via%20Embodied%20Prior%20Learning%20and%20Online%20Reinforcement%20Learning%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hanyang%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mark%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rui%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qinwei%22%2C%22lastName%22%3A%22Ma%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ke%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiarui%22%2C%22lastName%22%3A%22Yao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kangrui%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hao%22%2C%22lastName%22%3A%22Bai%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhenhailong%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rui%22%2C%22lastName%22%3A%22Pan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mengchao%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jose%22%2C%22lastName%22%3A%22Barreiros%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aykut%22%2C%22lastName%22%3A%22Onol%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22ChengXiang%22%2C%22lastName%22%3A%22Zhai%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Heng%22%2C%22lastName%22%3A%22Ji%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Manling%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Huan%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tong%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22Recent%20advances%20in%20embodied%20AI%20highlight%20the%20potential%20of%20vision%20language%20models%20%28VLMs%29%20as%20agents%20capable%20of%20perception%2C%20reasoning%2C%20and%20interaction%20in%20complex%20environments.%20However%2C%20top-performing%20systems%20rely%20on%20large-scale%20models%20that%20are%20costly%20to%20deploy%2C%20while%20smaller%20VLMs%20lack%20the%20necessary%20knowledge%20and%20skills%20to%20succeed.%20To%20bridge%20this%20gap%2C%20we%20present%20%5C%5Ctextit%7BEmbodied%20Reasoning%20Agent%20%28ERA%29%7D%2C%20a%20two-stage%20framework%20that%20integrates%20prior%20knowledge%20learning%20and%20online%20reinforcement%20learning%20%28RL%29.%20The%20first%20stage%2C%20%5C%5Ctextit%7BEmbodied%20Prior%20Learning%7D%2C%20distills%20foundational%20knowledge%20from%20three%20types%20of%20data%3A%20%281%29%20Trajectory-Augmented%20Priors%2C%20which%20enrich%20existing%20trajectory%20data%20with%20structured%20reasoning%20generated%20by%20stronger%20models%3B%20%282%29%20Environment-Anchored%20Priors%2C%20which%20provide%20in-environment%20knowledge%20and%20grounding%20supervision%3B%20and%20%283%29%20External%20Knowledge%20Priors%2C%20which%20transfer%20general%20knowledge%20from%20out-of-environment%20datasets.%20In%20the%20second%20stage%2C%20we%20develop%20an%20online%20RL%20pipeline%20that%20builds%20on%20these%20priors%20to%20further%20enhance%20agent%20performance.%20To%20overcome%20the%20inherent%20challenges%20in%20agent%20RL%2C%20including%20long%20horizons%2C%20sparse%20rewards%2C%20and%20training%20instability%2C%20we%20introduce%20three%20key%20designs%3A%20self-summarization%20for%20context%20management%2C%20dense%20reward%20shaping%2C%20and%20turn-level%20policy%20optimization.%20Extensive%20experiments%20on%20both%20high-level%20planning%20%28EB-ALFRED%29%20and%20low-level%20control%20%28EB-Manipulation%29%20tasks%20demonstrate%20that%20ERA-3B%20surpasses%20both%20prompting-based%20large%20models%20and%20previous%20training-based%20baselines.%20Specifically%2C%20it%20achieves%20overall%20improvements%20of%208.4%5C%5C%25%20on%20EB-ALFRED%20and%2019.4%5C%5C%25%20on%20EB-Manipulation%20over%20GPT-4o%2C%20and%20exhibits%20strong%20generalization%20to%20unseen%20tasks.%20Overall%2C%20ERA%20offers%20a%20practical%20path%20toward%20scalable%20embodied%20intelligence%2C%20providing%20methodological%20insights%20for%20future%20embodied%20AI%20systems.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.12693%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.12693%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T21%3A14%3A11Z%22%7D%7D%2C%7B%22key%22%3A%22XKBVDHPY%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wu%20and%20Zhang%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWu%2C%20M.%20%26amp%3B%20Zhang%2C%20Z.%20Maple%3A%20A%20Multi-agent%20System%20for%20Portable%20Deep%20Learning%20across%20Clusters.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.08842%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2510.08842%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Maple%3A%20A%20Multi-agent%20System%20for%20Portable%20Deep%20Learning%20across%20Clusters%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Molang%22%2C%22lastName%22%3A%22Wu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhao%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22Training%20deep%20learning%20%28DL%29%20models%20across%20Graphics%20Processing%20Unit%20%28GPU%29%20clusters%20is%20technically%20challenging.%20One%20aspect%20is%20that%20users%20have%20to%20compose%20command%20lines%20to%20adapt%20to%20the%20heterogeneous%20launchers%2C%20schedulers%2C%20affinity%20options%2C%20DL%20framework%20arguments%2C%20and%20environment%20variables.%20Composing%20correct%20command%20lines%20is%20error-prone%20and%20can%20easily%20frustrate%20users%2C%20impeding%20research%20or%20wasting%20resources.%20In%20this%20work%2C%20we%20present%20Maple%2C%20a%20multi-agent%20system%20that%20generates%20correct%20DL%20command%20lines%20with%20users%26%23039%3B%20natural%20language%20input.%20Maple%20consists%20of%20four%20agents%20with%20the%20functionalities%20of%20information%20extraction%2C%20template%20retrieval%2C%20command%20line%20verification%2C%20and%20error%20correction.%20We%20evaluate%20Maple%20on%20nine%20GPU%20clusters%20across%20national%20computing%20centers%20in%20the%20U.S.%2C%20five%20representative%20deep%20learning%20model%20families%2C%20and%20four%20commonly%20used%20parallel%20DL%20training%20paradigms.%20Our%20experiments%20also%20cover%20schedulers%20of%20SLURM%20and%20PBS%20and%20heterogeneous%20architectures%2C%20such%20as%20NVIDIA%20A100%5C%2FH200%20GPUs%20and%20Intel%20Max%20series%20GPUs.%20Maple%20achieves%2092.0%25%20accuracy%20in%20generating%20command%20lines%20across%20the%20567%20test%20cases.%20Leverage%20multiple%20language%20models%20with%20an%20aggregated%20size%20of%2010B%20parameters%2C%20Maple%20delivers%20comparable%20performance%20to%20the%20state-of-the-art%20models%20of%20GPT-5%2C%20Claude%2C%20and%20Gemini.%20Together%2C%20these%20results%20highlight%20Maple%26%23039%3Bs%20practical%20value%20in%20enabling%20portable%20and%20scalable%20distributed%20DL%20across%20heterogeneous%20HPC%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2510.08842%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2510.08842%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T21%3A00%3A15Z%22%7D%7D%2C%7B%22key%22%3A%22UN479BQA%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Xie%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-15%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BXie%2C%20H.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Diamond%3A%20Harnessing%20GPU%20Resources%20for%20Scientific%20Deep%20Learning.%20in%20%26lt%3Bi%26gt%3B2025%20IEEE%20International%20Conference%20on%20eScience%20%28eScience%29%26lt%3B%5C%2Fi%26gt%3B%20196%26%23x2013%3B204%20%28IEEE%2C%20Chicago%2C%20IL%2C%20USA%2C%202025%29.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FeScience65000.2025.00031%26%23039%3B%26gt%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FeScience65000.2025.00031%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Diamond%3A%20Harnessing%20GPU%20Resources%20for%20Scientific%20Deep%20Learning%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Haotian%22%2C%22lastName%22%3A%22Xie%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rohan%22%2C%22lastName%22%3A%22Marwaha%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minu%22%2C%22lastName%22%3A%22Mathew%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Song%22%2C%22lastName%22%3A%22Bian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Gengcong%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minghao%22%2C%22lastName%22%3A%22Yan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yadu%22%2C%22lastName%22%3A%22Babuji%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Owen%22%2C%22lastName%22%3A%22Price%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yinzhi%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Volodymyr%22%2C%22lastName%22%3A%22Kindratenko%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shivaram%22%2C%22lastName%22%3A%22Venkataraman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kyle%22%2C%22lastName%22%3A%22Chard%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ian%20T.%22%2C%22lastName%22%3A%22Foster%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhao%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%222025%20IEEE%20International%20Conference%20on%20eScience%20%28eScience%29%22%2C%22conferenceName%22%3A%222025%20IEEE%20International%20Conference%20on%20eScience%20%28eScience%29%22%2C%22date%22%3A%222025-9-15%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1109%5C%2FeScience65000.2025.00031%22%2C%22ISBN%22%3A%229798331591458%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fieeexplore.ieee.org%5C%2Fdocument%5C%2F11181545%5C%2F%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T20%3A47%3A53Z%22%7D%7D%2C%7B%22key%22%3A%225D9LRUXC%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Patel%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BPatel%2C%20P.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20RADAR-Radio%20Afterglow%20Detection%20and%20AI-driven%20Response%3A%20A%20Federated%20Framework%20for%20Gravitational%20Wave%20Event%20Follow-Up.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2507.14827%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2507.14827%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22RADAR-Radio%20Afterglow%20Detection%20and%20AI-driven%20Response%3A%20A%20Federated%20Framework%20for%20Gravitational%20Wave%20Event%20Follow-Up%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Parth%22%2C%22lastName%22%3A%22Patel%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alessandra%22%2C%22lastName%22%3A%22Corsi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22E.%20A.%22%2C%22lastName%22%3A%22Huerta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kara%22%2C%22lastName%22%3A%22Merfeld%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Victoria%22%2C%22lastName%22%3A%22Tiki%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zilinghan%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tekin%22%2C%22lastName%22%3A%22Bicer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kyle%22%2C%22lastName%22%3A%22Chard%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ryan%22%2C%22lastName%22%3A%22Chard%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ian%20T.%22%2C%22lastName%22%3A%22Foster%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Maxime%22%2C%22lastName%22%3A%22Gonthier%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Valerie%22%2C%22lastName%22%3A%22Hayot-Sasson%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hai%20Duc%22%2C%22lastName%22%3A%22Nguyen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Haochen%22%2C%22lastName%22%3A%22Pan%22%7D%5D%2C%22abstractNote%22%3A%22The%20landmark%20detection%20of%20both%20gravitational%20waves%20%28GWs%29%20and%20electromagnetic%20%28EM%29%20radiation%20from%20the%20binary%20neutron%20star%20merger%20GW170817%20has%20spurred%20efforts%20to%20streamline%20the%20follow-up%20of%20GW%20alerts%20in%20current%20and%20future%20observing%20runs%20of%20ground-based%20GW%20detectors.%20Within%20this%20context%2C%20the%20radio%20band%20of%20the%20EM%20spectrum%20presents%20unique%20challenges.%20Sensitive%20radio%20facilities%20capable%20of%20detecting%20the%20faint%20radio%20afterglow%20seen%20in%20GW170817%2C%20and%20with%20sufficient%20angular%20resolution%2C%20have%20small%20fields%20of%20view%20compared%20to%20typical%20GW%20localization%20areas.%20Additionally%2C%20theoretical%20models%20predict%20that%20the%20radio%20emission%20from%20binary%20neutron%20star%20mergers%20can%20evolve%20over%20weeks%20to%20years%2C%20necessitating%20long-term%20monitoring%20to%20probe%20the%20physics%20of%20the%20various%20post-merger%20ejecta%20components.%20These%20constraints%2C%20combined%20with%20limited%20radio%20observing%20resources%2C%20make%20the%20development%20of%20more%20coordinated%20follow-up%20strategies%20essential%20--%20especially%20as%20the%20next%20generation%20of%20GW%20detectors%20promise%20a%20dramatic%20increase%20in%20detection%20rates.%20Here%2C%20we%20present%20RADAR%2C%20a%20framework%20designed%20to%20address%20these%20challenges%20by%20promoting%20community-driven%20information%20sharing%2C%20federated%20data%20analysis%2C%20and%20system%20resilience%2C%20while%20integrating%20AI%20methods%20for%20both%20GW%20signal%20identification%20and%20radio%20data%20aggregation.%20We%20show%20that%20it%20is%20possible%20to%20preserve%20data%20rights%20while%20sharing%20models%20that%20can%20help%20design%20and%5C%2For%20update%20follow-up%20strategies.%20We%20demonstrate%20our%20approach%20through%20a%20case%20study%20of%20GW170817%2C%20and%20discuss%20future%20directions%20for%20refinement%20and%20broader%20application.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2507.14827%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2507.14827%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T20%3A26%3A29Z%22%7D%7D%2C%7B%22key%22%3A%226AT3CREW%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Kacmaz%20et%20al.%22%2C%22parsedDate%22%3A%222025-08-29%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BKacmaz%2C%20S.%2C%20Haas%2C%20R.%20%26amp%3B%20Huerta%2C%20E.%20A.%20Machine%20Learning-Driven%20Conservative-to-Primitive%20Conversion%20in%20Hybrid%20Piecewise%20Polytropic%20and%20Tabulated%20Equations%20of%20State.%20%26lt%3Bi%26gt%3BSymmetry%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B17%26lt%3B%5C%2Fb%26gt%3B%2C%201409%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Machine%20Learning-Driven%20Conservative-to-Primitive%20Conversion%20in%20Hybrid%20Piecewise%20Polytropic%20and%20Tabulated%20Equations%20of%20State%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Semih%22%2C%22lastName%22%3A%22Kacmaz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Roland%22%2C%22lastName%22%3A%22Haas%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22E.%20A.%22%2C%22lastName%22%3A%22Huerta%22%7D%5D%2C%22abstractNote%22%3A%22We%20present%20a%20novel%20machine%20learning%20%28ML%29-based%20method%20to%20accelerate%20conservative-to-primitive%20inversion%2C%20focusing%20on%20hybrid%20piecewise%20polytropic%20and%20tabulated%20equations%20of%20state.%20Traditional%20root-finding%20techniques%20are%20computationally%20expensive%2C%20particularly%20for%20large-scale%20relativistic%20hydrodynamics%20simulations.%20To%20address%20this%2C%20we%20employ%20feedforward%20neural%20networks%20%28NNC2PS%20and%20NNC2PL%29%2C%20trained%20in%20PyTorch%20%282.0%2B%29%20and%20optimized%20for%20GPU%20inference%20using%20NVIDIA%20TensorRT%20%288.4.1%29%2C%20achieving%20significant%20speedups%20with%20minimal%20accuracy%20loss.%20The%20NNC2PS%20model%20achieves%20L1%20and%20L%5Cu221e%20errors%20of%204.54%5Cu00d710%5Cu22127%20and%203.44%5Cu00d710%5Cu22126%2C%20respectively%2C%20while%20the%20NNC2PL%20model%20exhibits%20even%20lower%20error%20values.%20TensorRT%20optimization%20with%20mixed-precision%20deployment%20substantially%20accelerates%20performance%20compared%20to%20traditional%20root-finding%20methods.%20Specifically%2C%20the%20mixed-precision%20TensorRT%20engine%20for%20NNC2PS%20achieves%20inference%20speeds%20approximately%20400%20times%20faster%20than%20a%20traditional%20single-threaded%20CPU%20implementation%20for%20a%20dataset%20size%20of%201%2C000%2C000%20points.%20Ideal%20parallelization%20across%20an%20entire%20compute%20node%20in%20the%20Delta%20supercomputer%20%28dual%20AMD%2064-core%202.45%20GHz%20Milan%20processors%20and%208%20NVIDIA%20A100%20GPUs%20with%2040%20GB%20HBM2%20RAM%20and%20NVLink%29%20predicts%20a%2025-fold%20speedup%20for%20TensorRT%20over%20an%20optimally%20parallelized%20numerical%20method%20when%20processing%208%20million%20data%20points.%20Moreover%2C%20the%20ML%20method%20exhibits%20sub-linear%20scaling%20with%20increasing%20dataset%20sizes.%20We%20release%20the%20scientific%20software%20developed%2C%20enabling%20further%20validation%20and%20extension%20of%20our%20findings.%20By%20exploiting%20the%20underlying%20symmetries%20within%20the%20equation%20of%20state%2C%20these%20findings%20highlight%20the%20potential%20of%20ML%2C%20combined%20with%20GPU%20optimization%20and%20model%20quantization%2C%20to%20accelerate%20conservative-to-primitive%20inversion%20in%20relativistic%20hydrodynamics%20simulations.%22%2C%22date%22%3A%222025-08-29%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.3390%5C%2Fsym17091409%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fwww.mdpi.com%5C%2F2073-8994%5C%2F17%5C%2F9%5C%2F1409%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222073-8994%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T20%3A02%3A37Z%22%7D%7D%2C%7B%22key%22%3A%22QMIZXUIF%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zhu%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZhu%2C%20M.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Probing%20the%20Critical%20Point%20%28CritPt%29%20of%20AI%20Reasoning%3A%20a%20Frontier%20Physics%20Research%20Benchmark.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.26574%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.26574%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Probing%20the%20Critical%20Point%20%28CritPt%29%20of%20AI%20Reasoning%3A%20a%20Frontier%20Physics%20Research%20Benchmark%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minhui%22%2C%22lastName%22%3A%22Zhu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minyang%22%2C%22lastName%22%3A%22Tian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiaocheng%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tianci%22%2C%22lastName%22%3A%22Zhou%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lifan%22%2C%22lastName%22%3A%22Yuan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Penghao%22%2C%22lastName%22%3A%22Zhu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Eli%22%2C%22lastName%22%3A%22Chertkov%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shengyan%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yufeng%22%2C%22lastName%22%3A%22Du%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ziming%22%2C%22lastName%22%3A%22Ji%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Indranil%22%2C%22lastName%22%3A%22Das%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Junyi%22%2C%22lastName%22%3A%22Cao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiabin%22%2C%22lastName%22%3A%22Yu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Peixue%22%2C%22lastName%22%3A%22Wu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jinchen%22%2C%22lastName%22%3A%22He%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yifan%22%2C%22lastName%22%3A%22Su%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yikun%22%2C%22lastName%22%3A%22Jiang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yujie%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chang%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ze-Min%22%2C%22lastName%22%3A%22Huang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Weizhen%22%2C%22lastName%22%3A%22Jia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yunkai%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Farshid%22%2C%22lastName%22%3A%22Jafarpour%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yong%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xinan%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jessie%22%2C%22lastName%22%3A%22Shelton%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aaron%20W.%22%2C%22lastName%22%3A%22Young%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22John%22%2C%22lastName%22%3A%22Bartolotta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wenchao%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yue%22%2C%22lastName%22%3A%22Sun%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Anjun%22%2C%22lastName%22%3A%22Chu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Victor%22%2C%22lastName%22%3A%22Colussi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chris%22%2C%22lastName%22%3A%22Akers%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nathan%22%2C%22lastName%22%3A%22Brooks%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wenbo%22%2C%22lastName%22%3A%22Fu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jinchao%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marvin%22%2C%22lastName%22%3A%22Qi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Anqi%22%2C%22lastName%22%3A%22Mu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yubo%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Allen%22%2C%22lastName%22%3A%22Zang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yang%22%2C%22lastName%22%3A%22Lyu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Peizhi%22%2C%22lastName%22%3A%22Mai%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Christopher%22%2C%22lastName%22%3A%22Wilson%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xuefei%22%2C%22lastName%22%3A%22Guo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Juntai%22%2C%22lastName%22%3A%22Zhou%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daniel%22%2C%22lastName%22%3A%22Inafuku%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chi%22%2C%22lastName%22%3A%22Xue%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Luyu%22%2C%22lastName%22%3A%22Gao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ze%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ya%5Cu00efr%22%2C%22lastName%22%3A%22Hein%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yonatan%22%2C%22lastName%22%3A%22Kahn%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kevin%22%2C%22lastName%22%3A%22Zhou%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Di%22%2C%22lastName%22%3A%22Luo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22John%20Drew%22%2C%22lastName%22%3A%22Wilson%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jarrod%20T.%22%2C%22lastName%22%3A%22Reilly%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Dmytro%22%2C%22lastName%22%3A%22Bandak%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ofir%22%2C%22lastName%22%3A%22Press%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Liang%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xueying%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hao%22%2C%22lastName%22%3A%22Tong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nicolas%22%2C%22lastName%22%3A%22Chia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Eliu%22%2C%22lastName%22%3A%22Huerta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hao%22%2C%22lastName%22%3A%22Peng%22%7D%5D%2C%22abstractNote%22%3A%22While%20large%20language%20models%20%28LLMs%29%20with%20reasoning%20capabilities%20are%20progressing%20rapidly%20on%20high-school%20math%20competitions%20and%20coding%2C%20can%20they%20reason%20effectively%20through%20complex%2C%20open-ended%20challenges%20found%20in%20frontier%20physics%20research%3F%20And%20crucially%2C%20what%20kinds%20of%20reasoning%20tasks%20do%20physicists%20want%20LLMs%20to%20assist%20with%3F%20To%20address%20these%20questions%2C%20we%20present%20the%20CritPt%20%28Complex%20Research%20using%20Integrated%20Thinking%20-%20Physics%20Test%2C%20pronounced%20%26quot%3Bcritical%20point%26quot%3B%29%2C%20the%20first%20benchmark%20designed%20to%20test%20LLMs%20on%20unpublished%2C%20research-level%20reasoning%20tasks%20that%20broadly%20covers%20modern%20physics%20research%20areas%2C%20including%20condensed%20matter%2C%20quantum%20physics%2C%20atomic%2C%20molecular%20%26amp%3B%20optical%20physics%2C%20astrophysics%2C%20high%20energy%20physics%2C%20mathematical%20physics%2C%20statistical%20physics%2C%20nuclear%20physics%2C%20nonlinear%20dynamics%2C%20fluid%20dynamics%20and%20biophysics.%20CritPt%20consists%20of%2071%20composite%20research%20challenges%20designed%20to%20simulate%20full-scale%20research%20projects%20at%20the%20entry%20level%2C%20which%20are%20also%20decomposed%20to%20190%20simpler%20checkpoint%20tasks%20for%20more%20fine-grained%20insights.%20All%20problems%20are%20newly%20created%20by%2050%2B%20active%20physics%20researchers%20based%20on%20their%20own%20research.%20Every%20problem%20is%20hand-curated%20to%20admit%20a%20guess-resistant%20and%20machine-verifiable%20answer%20and%20is%20evaluated%20by%20an%20automated%20grading%20pipeline%20heavily%20customized%20for%20advanced%20physics-specific%20output%20formats.%20We%20find%20that%20while%20current%20state-of-the-art%20LLMs%20show%20early%20promise%20on%20isolated%20checkpoints%2C%20they%20remain%20far%20from%20being%20able%20to%20reliably%20solve%20full%20research-scale%20challenges%3A%20the%20best%20average%20accuracy%20among%20base%20models%20is%20only%205.7%25%2C%20achieved%20by%20GPT-5%20%28high%29%2C%20moderately%20rising%20to%20around%2010%25%20when%20equipped%20with%20coding%20tools.%20Through%20the%20realistic%20yet%20standardized%20evaluation%20offered%20by%20CritPt%2C%20we%20highlight%20a%20large%20disconnect%20between%20current%20model%20capabilities%20and%20realistic%20physics%20research%20demands%2C%20offering%20a%20foundation%20to%20guide%20the%20development%20of%20scientifically%20grounded%20AI%20tools.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.26574%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.26574%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T18%3A41%3A32Z%22%7D%7D%2C%7B%22key%22%3A%22S8CCAQFP%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lian%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLian%2C%20X.%2C%20Tanaka%2C%20M.%2C%20Ruwase%2C%20O.%20%26amp%3B%20Zhang%2C%20M.%20SuperOffload%3A%20Unleashing%20the%20Power%20of%20Large-Scale%20LLM%20Training%20on%20Superchips.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.21271%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.21271%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22SuperOffload%3A%20Unleashing%20the%20Power%20of%20Large-Scale%20LLM%20Training%20on%20Superchips%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xinyu%22%2C%22lastName%22%3A%22Lian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Masahiro%22%2C%22lastName%22%3A%22Tanaka%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Olatunji%22%2C%22lastName%22%3A%22Ruwase%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minjia%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22The%20emergence%20of%20Superchips%20represents%20a%20significant%20advancement%20in%20next-generation%20AI%20hardware.%20These%20Superchips%20employ%20a%20tightly%20coupled%20heterogeneous%20architecture%20that%20integrates%20GPU%20and%20CPU%20on%20the%20same%20package%2C%20which%20offers%20unprecedented%20computational%20power.%20However%2C%20there%20has%20been%20scant%20research%20investigating%20how%20LLM%20training%20benefits%20from%20this%20new%20architecture.%20In%20this%20work%2C%20for%20the%20first%20time%2C%20we%20study%20LLM%20training%20solutions%20based%20on%20offloading%20for%20Superchips.%20We%20observe%20important%20differences%20between%20Superchips%20and%20traditional%20loosely-coupled%20GPU-CPU%20architecture%2C%20which%20necessitate%20revisiting%20prevailing%20assumptions%20about%20offloading.%20Based%20on%20that%2C%20we%20present%20SuperOffload%2C%20a%20Superchip-centric%20offloading%20system%20that%20simultaneously%20uses%20Hopper%20GPU%2C%20Grace%20CPU%2C%20and%20NVLink-C2C%20interconnect%20more%20efficiently.%20SuperOffload%20accomplishes%20this%20via%20a%20combination%20of%20techniques%2C%20such%20as%20adaptive%20weight%20offloading%2C%20bucketization%20repartitioning%2C%20Superchip-aware%20casting%2C%20speculative%20execution%2C%20and%20a%20highly%20optimized%20Adam%20optimizer%20for%20Grace%20CPUs.%20Our%20evaluation%20of%20SuperOffload%20on%20NVIDIA%20GH200%20demonstrates%20up%20to%202.5x%20throughput%20improvement%20compared%20to%20state-of-the-art%20offloading-based%20systems%2C%20enabling%20training%20of%20up%20to%2025B%20model%20on%20a%20single%20Superchip%20while%20achieving%20high%20training%20throughput.%20We%20also%20extend%20SuperOffload%20with%20ZeRO-style%20data%20parallelism%20and%20DeepSpeed-Ulysses%20sequence%20parallelism%2C%20enabling%20training%20of%2013B%20model%20with%20sequence%20lengths%20up%20to%201%20million%20tokens%20on%208%20GH200%20while%20achieving%2055%25%20MFU.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.21271%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.21271%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-12-05T18%3A29%3A10Z%22%7D%7D%2C%7B%22key%22%3A%222QMHZ93B%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Gerogiannis%20et%20al.%22%2C%22parsedDate%22%3A%222025-10-18%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BGerogiannis%2C%20G.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20NetSparse%3A%20In-Network%20Acceleration%20of%20Distributed%20Sparse%20Kernels.%20in%20%26lt%3Bi%26gt%3B58th%20IEEE%5C%2FACM%20International%20Symposium%20on%20Microarchitecture%26lt%3B%5C%2Fi%26gt%3B%20%28ACM%2C%20Seoul%2C%20Republic%20of%20Korea%2C%202025%29.%20doi%3Ahttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1145%5C%2F3725843.3756076.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22NetSparse%3A%20In-Network%20Acceleration%20of%20Distributed%20Sparse%20Kernels%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22G%22%2C%22lastName%22%3A%22Gerogiannis%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22D%22%2C%22lastName%22%3A%22Merkouriadis%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22C%22%2C%22lastName%22%3A%22Block%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22A%22%2C%22lastName%22%3A%22Zulfiqar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22F%22%2C%22lastName%22%3A%22Tofalos%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22M%22%2C%22lastName%22%3A%22Shahbaz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22J%22%2C%22lastName%22%3A%22Torrellas%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%2258th%20IEEE%5C%2FACM%20International%20Symposium%20on%20Microarchitecture%22%2C%22conferenceName%22%3A%22MICRO%20%5Cu201925%22%2C%22date%22%3A%222025-10-18%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%22https%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1145%5C%2F3725843.3756076%22%2C%22ISBN%22%3A%22979-8-4007-1573-0%5C%2F25%5C%2F10%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-23T21%3A18%3A50Z%22%7D%7D%2C%7B%22key%22%3A%22QNHLK9FL%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wong%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWong%2C%20G.%20N.%2C%20Medeiros%2C%20L.%20%26amp%3B%20Stone%2C%20J.%20M.%20Mass%20Transport%2C%20Turbulent%20Mixing%2C%20and%20Inflow%20in%20Black%20Hole%20Accretion.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.14202%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.14202%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Mass%20Transport%2C%20Turbulent%20Mixing%2C%20and%20Inflow%20in%20Black%20Hole%20Accretion%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22George%20N.%22%2C%22lastName%22%3A%22Wong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lia%22%2C%22lastName%22%3A%22Medeiros%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22James%20M.%22%2C%22lastName%22%3A%22Stone%22%7D%5D%2C%22abstractNote%22%3A%22We%20investigate%20mass%20transport%2C%20mixing%2C%20and%20disk%20evolution%20in%20non-radiative%20black%20hole%20accretion%20flows%20using%20Lagrangian%20tracer%20particles%20embedded%20in%20general%20relativistic%20magnetohydrodynamics%20simulations.%20Our%20simulation%20suite%20spans%20magnetically%20arrested%20disk%20%28MAD%29%20and%20standard%20and%20normal%20evolution%20%28SANE%29%20states%20across%20a%20range%20of%20black%20hole%20spins.%20By%20tracking%20tracer%20trajectories%2C%20we%20directly%20measure%20both%20advective%20inflow%20and%20stochastic%20spreading%20of%20fluid%20elements.%20The%20tracer%20distributions%20are%20well%20described%20by%20a%20combination%20of%20coherent%20inward%20drift%20and%20Gaussian-like%20broadening%2C%20consistent%20with%20an%20advection-diffusion%20picture.%20MADs%20exhibit%20systematically%20faster%20inflow%20than%20SANEs%2C%20with%20retrograde%20flows%20showing%20the%20most%20rapid%20infall%3B%20the%20innermost%20stable%20circular%20orbit%20leaves%20little%20imprint%20in%20MADs%20but%20remains%20more%20visible%20in%20SANEs.%20Turbulent%20fluctuations%20drive%20strong%20radial%20dispersion%20in%20all%20cases%2C%20with%20a%20superdiffusive%20scaling%20of%20sigma%20~%20t%5E0.95%20in%20MADs%20and%20sigma%20~%20t%5E0.75%20in%20SANEs%20for%20high-spin%20prograde%20disks.%20Mixing%20times%20decrease%20toward%20the%20event%20horizon%20and%20are%20consistently%20shorter%20in%20MADs%20and%20retrograde%20configurations.%20Tracers%20also%20reveal%20how%20accretion%20sources%20shift%20over%20time%3A%20turbulence%20draws%20inflow%20from%20a%20broad%20range%20of%20initial%20radii%2C%20with%20rapid%20torus%20depletion%20in%20MADs%20driving%20the%20mean%20source%20radius%20outward%20as%20r%20~%20t%5E%282%5C%2F3%29%2C%20while%20SANEs%20evolve%20more%20gradually%20with%20r%20~%20t%5E%281%5C%2F2%29.%20We%20show%20that%20the%20finite%20mass%20of%20the%20initial%20torus%20has%20a%20strong%20influence%20on%20late-time%20behavior%2C%20especially%20in%20MADs%2C%20where%20imprints%20of%20differently%20sized%20initial%20conditions%20may%20be%20accessible%20as%20early%20as%20t%20~%2010000%20GM%5C%2Fc%5E3.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.14202%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.14202%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-23T21%3A00%3A46Z%22%7D%7D%2C%7B%22key%22%3A%22NTHBF4BY%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zhong%20and%20Most%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZhong%2C%20Y.%20%26amp%3B%20Most%2C%20E.%20R.%20Unraveling%20the%20emission%20mechanism%20powering%20long%20period%20radio%20transients%20from%20interacting%20white%20dwarf%20binaries%20via%20kinetic%20plasma%20simulations.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.09057%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.09057%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Unraveling%20the%20emission%20mechanism%20powering%20long%20period%20radio%20transients%20from%20interacting%20white%20dwarf%20binaries%20via%20kinetic%20plasma%20simulations%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yici%22%2C%22lastName%22%3A%22Zhong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Elias%20R.%22%2C%22lastName%22%3A%22Most%22%7D%5D%2C%22abstractNote%22%3A%22Recent%20observations%20of%20long%20period%20radio%20transients%2C%20such%20as%20GLEAM-X%20J0704-37%20and%20ILTJ1101%20%2B%205521%2C%20have%20revealed%20a%20previously%20unrecognized%20population%20of%20galactic%20radio%20transient%20sources%20associated%20with%20white%20dwarf%20-%20M%20dwarf%20binaries.%20It%20is%20an%20open%20question%20how%20to%20produce%20coherent%20radio%20emission%20in%20these%20systems%2C%20though%20a%20model%20driven%20by%20binary%20interaction%20seems%20likely%20given%20the%20nature%20and%20correlation%20of%20the%20emission%20with%20the%20binaries%26%23039%3B%20orbital%20period.%20Using%20kinetic%20plasma%20simulations%2C%20we%20demonstrate%20that%20the%20relativistic%20electron%20cyclotron%20maser%20instability%20%28ECMI%29%20is%20a%20viable%20mechanism%20for%20generating%20radio%20pulses%20in%20white%20dwarf%20-%20M%20dwarf%20systems%2C%20akin%20to%20planetary%20radio%20emission%2C%20such%20as%20that%20from%20the%20Jupiter-Io%20system.%20We%20quantify%20the%20relativistic%20ECMI%20in%20the%20nonlinear%20regime%20under%20conditions%20relevant%20for%20white%20dwarf%20radio%20emission%20for%20the%20first%20time.%20Our%20simulations%20demonstrate%20that%20the%20ECMI%20can%20intrinsically%20produce%20partially%20linearly%20polarized%20emission%20relevant%20to%20explaining%20the%20observed%20emission%20spectrum%20of%20the%20two%20galactic%20sources%2C%20though%20the%20precise%20details%20will%20depend%20on%20the%20plasma%20composition.%20Our%20work%20paves%20the%20way%20for%20a%20systematic%20and%20fully%20nonlinear%20computational%20modeling%20of%20radio%20emission%20from%20interacting%20white%20dwarf%20sources.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.09057%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.09057%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A41%3A10Z%22%7D%7D%2C%7B%22key%22%3A%22AGUA7EZW%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Agrawalla%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BAgrawalla%2C%20B.%2C%20Nauman%2C%20M.%2C%20Agarwal%2C%20K.%20%26amp%3B%20Kumar%2C%20A.%20floq%3A%20Training%20Critics%20via%20Flow-Matching%20for%20Scaling%20Compute%20in%20Value-Based%20RL.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.06863%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.06863%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22floq%3A%20Training%20Critics%20via%20Flow-Matching%20for%20Scaling%20Compute%20in%20Value-Based%20RL%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bhavya%22%2C%22lastName%22%3A%22Agrawalla%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michal%22%2C%22lastName%22%3A%22Nauman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Khush%22%2C%22lastName%22%3A%22Agarwal%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aviral%22%2C%22lastName%22%3A%22Kumar%22%7D%5D%2C%22abstractNote%22%3A%22A%20hallmark%20of%20modern%20large-scale%20machine%20learning%20techniques%20is%20the%20use%20of%20training%20objectives%20that%20provide%20dense%20supervision%20to%20intermediate%20computations%2C%20such%20as%20teacher%20forcing%20the%20next%20token%20in%20language%20models%20or%20denoising%20step-by-step%20in%20diffusion%20models.%20This%20enables%20models%20to%20learn%20complex%20functions%20in%20a%20generalizable%20manner.%20Motivated%20by%20this%20observation%2C%20we%20investigate%20the%20benefits%20of%20iterative%20computation%20for%20temporal%20difference%20%28TD%29%20methods%20in%20reinforcement%20learning%20%28RL%29.%20Typically%20they%20represent%20value%20functions%20in%20a%20monolithic%20fashion%2C%20without%20iterative%20compute.%20We%20introduce%20floq%20%28flow-matching%20Q-functions%29%2C%20an%20approach%20that%20parameterizes%20the%20Q-function%20using%20a%20velocity%20field%20and%20trains%20it%20using%20techniques%20from%20flow-matching%2C%20typically%20used%20in%20generative%20modeling.%20This%20velocity%20field%20underneath%20the%20flow%20is%20trained%20using%20a%20TD-learning%20objective%2C%20which%20bootstraps%20from%20values%20produced%20by%20a%20target%20velocity%20field%2C%20computed%20by%20running%20multiple%20steps%20of%20numerical%20integration.%20Crucially%2C%20floq%20allows%20for%20more%20fine-grained%20control%20and%20scaling%20of%20the%20Q-function%20capacity%20than%20monolithic%20architectures%2C%20by%20appropriately%20setting%20the%20number%20of%20integration%20steps.%20Across%20a%20suite%20of%20challenging%20offline%20RL%20benchmarks%20and%20online%20fine-tuning%20tasks%2C%20floq%20improves%20performance%20by%20nearly%201.8x.%20floq%20scales%20capacity%20far%20better%20than%20standard%20TD-learning%20architectures%2C%20highlighting%20the%20potential%20of%20iterative%20computation%20for%20value%20learning.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.06863%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.06863%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A39%3A21Z%22%7D%7D%2C%7B%22key%22%3A%22RXQK7FEP%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Jafari%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BJafari%2C%20S.%2C%20Sarkar%2C%20A.%2C%20Bilwal%2C%20M.%20%26amp%3B%20Jannesari%2C%20A.%20ProfilingAgent%3A%20Profiling-Guided%20Agentic%20Reasoning%20for%20Adaptive%20Model%20Optimization.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.05584%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.05584%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22ProfilingAgent%3A%20Profiling-Guided%20Agentic%20Reasoning%20for%20Adaptive%20Model%20Optimization%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sadegh%22%2C%22lastName%22%3A%22Jafari%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aishwarya%22%2C%22lastName%22%3A%22Sarkar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mohiuddin%22%2C%22lastName%22%3A%22Bilwal%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ali%22%2C%22lastName%22%3A%22Jannesari%22%7D%5D%2C%22abstractNote%22%3A%22Foundation%20models%20face%20growing%20compute%20and%20memory%20bottlenecks%2C%20hindering%20deployment%20on%20resource-limited%20platforms.%20While%20compression%20techniques%20such%20as%20pruning%20and%20quantization%20are%20widely%20used%2C%20most%20rely%20on%20uniform%20heuristics%20that%20ignore%20architectural%20and%20runtime%20heterogeneity.%20Profiling%20tools%20expose%20per-layer%20latency%2C%20memory%2C%20and%20compute%20cost%2C%20yet%20are%20rarely%20integrated%20into%20automated%20pipelines.%20We%20propose%20ProfilingAgent%2C%20a%20profiling-guided%2C%20agentic%20approach%20that%20uses%20large%20language%20models%20%28LLMs%29%20to%20automate%20compression%20via%20structured%20pruning%20and%20post-training%20dynamic%20quantization.%20Our%20modular%20multi-agent%20system%20reasons%20over%20static%20metrics%20%28MACs%2C%20parameter%20counts%29%20and%20dynamic%20signals%20%28latency%2C%20memory%29%20to%20design%20architecture-specific%20strategies.%20Unlike%20heuristic%20baselines%2C%20ProfilingAgent%20tailors%20layer-wise%20decisions%20to%20bottlenecks.%20Experiments%20on%20ImageNet-1K%2C%20CIFAR-10%2C%20and%20CIFAR-100%20with%20ResNet-101%2C%20ViT-B%5C%2F16%2C%20Swin-B%2C%20and%20DeiT-B%5C%2F16%20show%20pruning%20maintains%20competitive%20or%20improved%20accuracy%20%28about%201%25%20drop%20on%20ImageNet-1K%2C%20%2B2%25%20gains%20for%20ViT-B%5C%2F16%20on%20smaller%20datasets%29%2C%20while%20quantization%20achieves%20up%20to%2074%25%20memory%20savings%20with%20%26lt%3B0.5%25%20accuracy%20loss.%20Our%20quantization%20also%20yields%20consistent%20inference%20speedups%20of%20up%20to%201.74%20times%20faster.%20Comparative%20studies%20with%20GPT-4o%20and%20GPT-4-Turbo%20highlight%20the%20importance%20of%20LLM%20reasoning%20quality%20for%20iterative%20pruning.%20These%20results%20establish%20agentic%20systems%20as%20scalable%20solutions%20for%20profiling-guided%20model%20optimization.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.05584%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.05584%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A38%3A03Z%22%7D%7D%2C%7B%22key%22%3A%22GL89S4GP%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wilfong%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWilfong%2C%20B.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Testing%20and%20benchmarking%20emerging%20supercomputers%20via%20the%20MFC%20flow%20solver.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.13575%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.13575%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Testing%20and%20benchmarking%20emerging%20supercomputers%20via%20the%20MFC%20flow%20solver%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Benjamin%22%2C%22lastName%22%3A%22Wilfong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Anand%22%2C%22lastName%22%3A%22Radhakrishnan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Henry%20A.%20Le%22%2C%22lastName%22%3A%22Berre%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tanush%22%2C%22lastName%22%3A%22Prathi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Stephen%22%2C%22lastName%22%3A%22Abbott%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Spencer%20H.%22%2C%22lastName%22%3A%22Bryngelson%22%7D%5D%2C%22abstractNote%22%3A%22Deploying%20new%20supercomputers%20requires%20testing%20and%20evaluation%20via%20application%20codes.%20Portable%2C%20user-friendly%20tools%20enable%20evaluation%2C%20and%20the%20Multicomponent%20Flow%20Code%20%28MFC%29%2C%20a%20computational%20fluid%20dynamics%20%28CFD%29%20code%2C%20addresses%20this%20need.%20MFC%20is%20adorned%20with%20a%20toolchain%20that%20automates%20input%20generation%2C%20compilation%2C%20batch%20job%20submission%2C%20regression%20testing%2C%20and%20benchmarking.%20The%20toolchain%20design%20enables%20users%20to%20evaluate%20compiler-hardware%20combinations%20for%20correctness%20and%20performance%20with%20limited%20software%20engineering%20experience.%20As%20with%20other%20PDE%20solvers%2C%20wall%20time%20per%20spatially%20discretized%20grid%20point%20serves%20as%20a%20figure%20of%20merit.%20We%20present%20MFC%20benchmarking%20results%20for%20five%20generations%20of%20NVIDIA%20GPUs%2C%20three%20generations%20of%20AMD%20GPUs%2C%20and%20various%20CPU%20architectures%2C%20utilizing%20Intel%2C%20Cray%2C%20NVIDIA%2C%20AMD%2C%20and%20GNU%20compilers.%20These%20tests%20have%20revealed%20compiler%20bugs%20and%20regressions%20on%20recent%20machines%20such%20as%20Frontier%20and%20El%20Capitan.%20MFC%20has%20benchmarked%20approximately%2050%20compute%20devices%20and%205%20flagship%20supercomputers.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.13575%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.13575%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A35%3A59Z%22%7D%7D%2C%7B%22key%22%3A%228FEEWMH6%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Bazavov%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BBazavov%2C%20A.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20High-Precision%20Scale%20Setting%20with%20the%20Omega-Baryon%20Mass%20and%20Gradient%20Flow.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.14367%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.14367%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22High-Precision%20Scale%20Setting%20with%20the%20Omega-Baryon%20Mass%20and%20Gradient%20Flow%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alexei%22%2C%22lastName%22%3A%22Bazavov%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Claude%20W.%22%2C%22lastName%22%3A%22Bernard%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22David%20A.%22%2C%22lastName%22%3A%22Clarke%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Carleton%22%2C%22lastName%22%3A%22DeTar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aida%20X.%22%2C%22lastName%22%3A%22El-Khadra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Elvira%22%2C%22lastName%22%3A%22G%5Cu00e1miz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Steven%22%2C%22lastName%22%3A%22Gottlieb%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Anthony%20V.%22%2C%22lastName%22%3A%22Grebe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Urs%20M.%22%2C%22lastName%22%3A%22Heller%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Leon%22%2C%22lastName%22%3A%22Hostetler%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%20I.%22%2C%22lastName%22%3A%22Jay%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hwancheol%22%2C%22lastName%22%3A%22Jeong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andreas%20S.%22%2C%22lastName%22%3A%22Kronfeld%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yin%22%2C%22lastName%22%3A%22Lin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shaun%22%2C%22lastName%22%3A%22Lahert%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jack%22%2C%22lastName%22%3A%22Laiho%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%22%2C%22lastName%22%3A%22Lynch%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andrew%20T.%22%2C%22lastName%22%3A%22Lytle%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aaron%20S.%22%2C%22lastName%22%3A%22Meyer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ethan%20T.%22%2C%22lastName%22%3A%22Neil%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Curtis%20T.%22%2C%22lastName%22%3A%22Peterson%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22James%20N.%22%2C%22lastName%22%3A%22Simone%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jacob%20W.%22%2C%22lastName%22%3A%22Sitison%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ruth%20S.%22%2C%22lastName%22%3A%22Van%20de%20Water%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alejandro%22%2C%22lastName%22%3A%22Vaquero%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%20L.%22%2C%22lastName%22%3A%22Wagman%22%7D%5D%2C%22abstractNote%22%3A%22The%20gradient-flow%20scale%20%24w_0%24%20in%20lattice%20QCD%20is%20determined%20using%20the%20mass%20of%20the%20%24%5Cu03a9%5E-%24%20baryon%20to%20set%20the%20physical%20scale.%20Nine%20ensembles%20using%20the%20highly%20improved%20staggered%20quark%20%28HISQ%29%20action%20with%20lattice%20spacings%20of%200.15%20fm%20down%20to%200.04%20fm%20are%20used%2C%20seven%20of%20which%20have%20nearly%20physical%20light-quark%20masses.%20Electromagnetic%20corrections%20to%20the%20%24%5Cu03a9%5E-%24%20mass%20are%20defined%20in%20order%20to%20compute%20a%20pure-QCD%20%24%5Cu03a9%24%20mass.%20The%20final%20result%20is%20%24w_0%20%3D%200.17187%2868%29%24%20fm%2C%20corresponding%20to%20a%20relative%20uncertainty%20of%200.40%25%20and%20a%20central%20value%20in%20good%20agreement%20with%20previous%20calculations%20in%20the%20literature.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.14367%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.14367%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A33%3A37Z%22%7D%7D%2C%7B%22key%22%3A%22QGZP9HXR%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Yu%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BYu%2C%20J.%2C%20Taneja%2C%20A.%2C%20Lin%2C%20J.%20%26amp%3B%20Zhang%2C%20M.%20VoltanaLLM%3A%20Feedback-Driven%20Frequency%20Control%20and%20State-Space%20Routing%20for%20Energy-Efficient%20LLM%20Serving.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.04827%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.04827%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22VoltanaLLM%3A%20Feedback-Driven%20Frequency%20Control%20and%20State-Space%20Routing%20for%20Energy-Efficient%20LLM%20Serving%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiahuan%22%2C%22lastName%22%3A%22Yu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aryan%22%2C%22lastName%22%3A%22Taneja%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Junfeng%22%2C%22lastName%22%3A%22Lin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minjia%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22Modern%20Large%20Language%20Model%20%28LLM%29%20serving%20systems%20increasingly%20support%20interactive%20applications%2C%20like%20real-time%20chat%20assistants%2C%20code%20generation%20tools%2C%20and%20agentic%20workflows.%20However%2C%20the%20soaring%20energy%20cost%20of%20LLM%20inference%20presents%20a%20growing%20challenge%20for%20sustainable%20and%20cost-effective%20deployment.%20This%20paper%20introduces%20VoltanaLLM%2C%20a%20system%20for%20SLO-aware%2C%20energy-efficient%20LLM%20serving%2C%20built%20from%20a%20control%20theory%20perspective.%20VoltanaLLM%20co-designs%20frequency%20scaling%20and%20request%20routing%20in%20emerging%20prefill%5C%2Fdecode%20disaggregated%20architectures%2C%20leveraging%20their%20decoupled%20execution%20to%20enable%20fine-grained%20phase-specific%20control.%20It%20consists%20of%20a%20feedback-driven%20frequency%20controller%20that%20dynamically%20adapts%20GPU%20frequency%20for%20prefill%20and%20decode%20phases%2C%20and%20a%20state-space%20router%20that%20explores%20routing%20decisions%20across%20frequency-scaled%20instances%20to%20minimize%20energy%20under%20latency%20constraints.%20We%20implement%20VoltanaLLM%20in%20SGLang%20and%20evaluate%20its%20performance%20over%20multiple%20state-of-the-art%20LLMs%20and%20real-world%20datasets.%20The%20results%20demonstrate%20that%20VoltanaLLM%20achieves%20up%20to%2036.3%25%20energy%20savings%20while%20maintaining%20near-perfect%20SLO%20attainment%20rate%2C%20paving%20the%20way%20for%20sustainable%20and%20intelligent%20LLM%20serving.%20Code%20of%20VoltanaLLM%20is%20open-sourced%20on%20GitHub%3A%20https%3A%5C%2F%5C%2Fgithub.com%5C%2FSupercomputing-System-AI-Lab%5C%2FVoltanaLLM.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.04827%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.04827%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A20%3A06Z%22%7D%7D%2C%7B%22key%22%3A%22ET3FBEG5%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wang%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-05%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWang%2C%20S.%2C%20Nguyen%2C%20H.%2C%20Eick%2C%20B.%2C%20Wilson%2C%20R.%20%26amp%3B%20El-Gohary%2C%20N.%20Efficient%20domain%20adaptation%20network%20for%20corrosion%20segmentation%20in%20infrastructure%3A%20from%20laboratory%20to%20real-world%20images.%20%26lt%3Bi%26gt%3BJ%20Civil%20Struct%20Health%20Monit%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2Fs13349-025-01003-y%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2Fs13349-025-01003-y%26lt%3B%5C%2Fa%26gt%3B%20%282025%29%20doi%3A10.1007%5C%2Fs13349-025-01003-y.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Efficient%20domain%20adaptation%20network%20for%20corrosion%20segmentation%20in%20infrastructure%3A%20from%20laboratory%20to%20real-world%20images%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shengyi%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hai%22%2C%22lastName%22%3A%22Nguyen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Brian%22%2C%22lastName%22%3A%22Eick%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rebekah%22%2C%22lastName%22%3A%22Wilson%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nora%22%2C%22lastName%22%3A%22El-Gohary%22%7D%5D%2C%22abstractNote%22%3A%22Abstract%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Corrosion%20in%20infrastructures%2C%20such%20as%20locks%2C%20dams%2C%20and%20facilities%2C%20poses%20significant%20risks%20to%20structural%20integrity%20and%20drives%20up%20maintenance%20costs.%20Accurate%20detection%20and%20segmentation%20of%20corrosion%20from%20inspection%20imagery%20are%20critical%20for%20proactive%20structural%20health%20monitoring%20%28SHM%29%20and%20maintenance%20decision%20making.%20Traditional%20methods%20for%20corrosion%20assessment%20rely%20heavily%20on%20manual%20inspection%2C%20which%20are%20time-consuming%2C%20inconsistent%2C%20and%20often%20ineffective%20under%20diverse%20environmental%20conditions.%20Recent%20advancements%20in%20deep%20learning%20have%20improved%20automated%20image-based%20corrosion%20detection.%20However%2C%20many%20existing%20approaches%20struggle%20with%20generalization%20across%20domains%20due%20to%20variability%20in%20corrosion%20appearance%2C%20surface%20materials%2C%20and%20imaging%20conditions%20%28e.g.%2C%20laboratory%20versus%20field%29.%20To%20address%20this%20challenge%2C%20this%20paper%20proposes%20a%20deep%20learning-based%20corrosion%20detection%20framework%20that%20integrates%20domain%20adaptation%20to%20enable%20robust%20segmentation%20performance%20across%20heterogeneous%20datasets.%20The%20proposed%20method%20consists%20of%20four%20key%20components%3A%20%281%29%20a%20geometric%20structure%20preservation%20%28GSP%29%20module%20to%20retain%20local%20topological%20relationships%20across%20domains%2C%20%282%29%20a%20singular%20value%20decomposition%20%28SVD%29-based%20local%20discrepancy%20%28SLD%29%20module%20that%20leverages%20SVD%20to%20align%20local%20feature%20subspaces%20at%20a%20fine-grained%20level%2C%20%283%29%20a%20global%20consistency%20alignment%20%28GCA%29%20module%20using%20maximum%20mean%20idscrepancy%20%28MMD%29%20to%20reconcile%20distributional%20shifts%20between%20source%20and%20target%20domains%2C%20and%20%284%29%20an%20auxiliary%20domain%20adversarial%20neural%20network%20%28DANN%29%20component%20to%20further%20encourage%20domain-invariant%20feature%20learning.%20These%20modules%20are%20supported%20by%20a%20proposed%20lightweight%20segmentation%20backbone%2C%20the%20efficient%20segmentation%20network%20%28EffSegNet%29%2C%20which%20enables%20efficient%20corrosion%20pattern%20recognition.%20The%20proposed%20framework%20was%20implemented%20and%20tested%20on%20corrosion%20imagery%20from%20real-world%20infrastructure%20assets.%20The%20method%20achieved%20high%20segmentation%20accuracy%20and%20demonstrated%20strong%20generalization%20capabilities%20across%20domain-shifted%20datasets%2C%20indicating%20its%20potential%20to%20support%20scalable%2C%20efficient%2C%20and%20automated%20corrosion%20assessment%20for%20SHM%20applications.%22%2C%22date%22%3A%222025-09-05%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1007%5C%2Fs13349-025-01003-y%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Flink.springer.com%5C%2F10.1007%5C%2Fs13349-025-01003-y%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222190-5452%2C%202190-5479%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A13%3A41Z%22%7D%7D%2C%7B%22key%22%3A%22M4PD3KIH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Bhatia%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BBhatia%2C%20M.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20%5C%5Ctextit%7BIn%20Silico%7D%20Benchmarking%20of%20Detectable%20Byzantine%20Agreement%20in%20Noisy%20Quantum%20Networks.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.02629%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.02629%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22%5C%5Ctextit%7BIn%20Silico%7D%20Benchmarking%20of%20Detectable%20Byzantine%20Agreement%20in%20Noisy%20Quantum%20Networks%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mayank%22%2C%22lastName%22%3A%22Bhatia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shaan%22%2C%22lastName%22%3A%22Doshi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daniel%22%2C%22lastName%22%3A%22Winton%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Brian%22%2C%22lastName%22%3A%22Doolittle%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bruno%22%2C%22lastName%22%3A%22Abreu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Santiago%22%2C%22lastName%22%3A%22N%5Cu00fa%5Cu00f1ez-Corrales%22%7D%5D%2C%22abstractNote%22%3A%22Quantum%20communication%20resources%20offer%20significant%20advantages%20for%20fault-tolerant%20distributed%20protocols%2C%20particularly%20in%20Byzantine%20Agreement%20%28BA%29%2C%20where%20reliability%20against%20adversarial%20interference%20is%20essential.%20Quantum%20Detectable%20Byzantine%20Agreement%20%28QDBA%29%20enables%20consensus%20protocols%20that%20surpass%20classical%20limitations%20by%20leveraging%20entangled%20quantum%20states.%20In%20this%20work%2C%20we%20focus%20on%20the%20practical%20realization%20of%20QDBA%20using%20Einstein-Podolsky-Rosen%20%28EPR%29%20pairs%2C%20the%20simplest%20maximally%20entangled%20quantum%20resources%2C%20making%20the%20protocol%20experimentally%20accessible%20across%20current%20quantum%20hardware%20platforms.%20We%20present%20a%20comprehensive%20computational%20study%20of%20the%20EPRQDBA%20protocol%20under%20realistic%20quantum%20network%20conditions%2C%20utilizing%20the%20Aliro%20Quantum%20Network%20Simulator%20to%20evaluate%20the%20performance%20and%20robustness%20of%20the%20protocol.%20Our%20simulations%20systematically%20explore%20the%20protocol%26%23039%3Bs%20parameter%20space%20--including%20variations%20in%20network%20size%2C%20traitorous%20node%20count%2C%20the%20amount%20of%20entanglement%20consumed%20in%20the%20protocol%2C%20and%20physically%20motivated%20noise%20models%20tailored%20specifically%20for%20superconducting%20and%20photonic%20qubit%20technologies.%20Through%20extensive%20numerical%20experiments%2C%20we%20provide%20insights%20into%20how%20these%20physically%20realistic%20parameters%20impact%20protocol%20performance%2C%20establishing%20critical%20thresholds%20and%20optimal%20operational%20regimes%20for%20experimental%20implementations.%20This%20work%20bridges%20theoretical%20advances%20in%20quantum%20consensus%20protocols%20with%20practical%20network%20implementations%2C%20offering%20a%20concrete%20reference%20for%20experimentalists.%20Our%20findings%20serve%20as%20a%20guideline%20for%20evaluating%20and%20optimizing%20QDBA%20implementations%20in%20realistic%2C%20noisy%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.02629%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.02629%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T22%3A03%3A44Z%22%7D%7D%2C%7B%22key%22%3A%22MXWBENXL%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Chantraine%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-05%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BChantraine%2C%20C.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Ultrastrong%20%26lt%3Bi%26gt%3BStaphylococcus%20aureus%26lt%3B%5C%2Fi%26gt%3B%20adhesion%20to%20human%20skin%3A%20Calcium%20as%20a%20key%20regulator%20of%20noncovalent%20interactions.%20%26lt%3Bi%26gt%3BSci.%20Adv.%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B11%26lt%3B%5C%2Fb%26gt%3B%2C%20eadu7457%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Ultrastrong%20%3Ci%3EStaphylococcus%20aureus%3C%5C%2Fi%3E%20adhesion%20to%20human%20skin%3A%20Calcium%20as%20a%20key%20regulator%20of%20noncovalent%20interactions%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Constance%22%2C%22lastName%22%3A%22Chantraine%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Priscila%20S.%20F.%20C.%22%2C%22lastName%22%3A%22Gomes%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marion%22%2C%22lastName%22%3A%22Matheli%5Cu00e9-Guinlet%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Diego%20E.%20B.%22%2C%22lastName%22%3A%22Gomes%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhiyong%22%2C%22lastName%22%3A%22Zheng%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Julianne%22%2C%22lastName%22%3A%22Clowry%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mary%20B.%22%2C%22lastName%22%3A%22Turley%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alan%20D.%22%2C%22lastName%22%3A%22Irvine%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Joan%20A.%22%2C%22lastName%22%3A%22Geoghegan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rafael%20C.%22%2C%22lastName%22%3A%22Bernardi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yves%20F.%22%2C%22lastName%22%3A%22Dufr%5Cu00eane%22%7D%5D%2C%22abstractNote%22%3A%22Calcium%20is%20a%20critical%20regulator%20of%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Staphylococcus%20aureus%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20skin%20adhesion%2C%20stabilizing%20one%20of%20the%20strongest%20noncovalent%20biomolecular%20interactions%20ever%20recorded.%20Using%20in%20vitro%20and%20in%20silico%20single-molecule%20force%20spectroscopy%2C%20we%20demonstrate%20that%20calcium%20ions%20%28Ca%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%202%2B%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%29%20are%20essential%20for%20the%20ultrastrong%20binding%20between%20the%20serine-aspartate%20repeat%20protein%20D%20%28SdrD%29%20adhesin%20and%20the%20human%20skin%20protein%20desmoglein-1%20%28DSG-1%29%2C%20withstanding%20forces%20exceeding%202%20nanonewtons.%20Ca%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%202%2B%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20ions%20stabilize%20both%20the%20SdrD%20complex%20and%20the%20mechanically%20robust%20SdrD%20B-domains%2C%20which%20exhibit%20unprecedented%20folding%20strength.%20In%20the%20context%20of%20atopic%20dermatitis%20%28AD%29%2C%20disrupted%20calcium%20gradients%20amplify%20SdrD%20interactions%2C%20which%20could%20potentially%20intensify%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20S.%20aureus%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20virulence.%20Furthermore%2C%20abnormal%20DSG-1%20distribution%20on%20AD-affected%20skin%20enhances%20bacterial%20adhesion.%20These%20findings%20provide%20crucial%20insights%20into%20the%20calcium-dependent%20regulation%20of%20bacterial%20adhesion%20and%20folding%2C%20suggesting%20possible%20therapeutic%20targets%20to%20combat%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20S.%20aureus%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20infections.%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%2C%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Calcium%20ions%20stabilize%20one%20of%20the%20strongest%20known%20noncovalent%20biological%20interactions.%22%2C%22date%22%3A%222025-09-05%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1126%5C%2Fsciadv.adu7457%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fwww.science.org%5C%2Fdoi%5C%2F10.1126%5C%2Fsciadv.adu7457%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222375-2548%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T14%3A19%3A08Z%22%7D%7D%2C%7B%22key%22%3A%223FPEH47M%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wang%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWang%2C%20P.%2C%20Watanabe%2C%20S.%20%26amp%3B%20Van%20hamme%2C%20H.%20SSVD%3A%20Structured%20SVD%20for%20Parameter-Efficient%20Fine-Tuning%20and%20Benchmarking%20under%20Domain%20Shift%20in%20ASR.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.02830%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2509.02830%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22SSVD%3A%20Structured%20SVD%20for%20Parameter-Efficient%20Fine-Tuning%20and%20Benchmarking%20under%20Domain%20Shift%20in%20ASR%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Pu%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shinji%22%2C%22lastName%22%3A%22Watanabe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hugo%22%2C%22lastName%22%3A%22Van%20hamme%22%7D%5D%2C%22abstractNote%22%3A%22Parameter-efficient%20fine-tuning%20%28PEFT%29%20has%20emerged%20as%20a%20scalable%20solution%20for%20adapting%20large%20foundation%20models.%20While%20low-rank%20adaptation%20%28LoRA%29%20is%20widely%20used%20in%20speech%20applications%2C%20its%20state-of-the-art%20variants%2C%20e.g.%2C%20VeRA%2C%20DoRA%2C%20PiSSA%2C%20and%20SVFT%2C%20are%20developed%20mainly%20for%20language%20and%20vision%20tasks%2C%20with%20limited%20validation%20in%20speech.%20This%20work%20presents%20the%20first%20comprehensive%20integration%20and%20benchmarking%20of%20these%20PEFT%20methods%20within%20ESPnet.%20We%20further%20introduce%20structured%20SVD-guided%20%28SSVD%29%20fine-tuning%2C%20which%20selectively%20rotates%20input-associated%20right%20singular%20vectors%20while%20keeping%20output-associated%20vectors%20fixed%20to%20preserve%20semantic%20mappings.%20This%20design%20enables%20robust%20domain%20adaptation%20with%20minimal%20trainable%20parameters%20and%20improved%20efficiency.%20We%20evaluate%20all%20methods%20on%20domain-shifted%20speech%20recognition%20tasks%2C%20including%20child%20speech%20and%20dialectal%20variation%2C%20across%20model%20scales%20from%200.1B%20to%202B.%20All%20implementations%20are%20released%20in%20ESPnet%20to%20support%20reproducibility%20and%20future%20work.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2509.02830%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2509.02830%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T14%3A16%3A22Z%22%7D%7D%2C%7B%22key%22%3A%223VQU6FM5%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Cho%20and%20Hockenmaier%22%2C%22parsedDate%22%3A%222025-08-26%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCho%2C%20I.%20%26amp%3B%20Hockenmaier%2C%20J.%20Analyzing%20Multilingualism%20in%20Large%20Language%20Models%20with%20Sparse%20Autoencoders.%20in%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Analyzing%20Multilingualism%20in%20Large%20Language%20Models%20with%20Sparse%20Autoencoders%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ikhyun%22%2C%22lastName%22%3A%22Cho%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Julia%22%2C%22lastName%22%3A%22Hockenmaier%22%7D%5D%2C%22abstractNote%22%3A%22Despite%20the%20impressive%20multilingual%20capabilities%20of%20recent%20large%20language%20models%20%28LLMs%29%2C%20the%20mechanisms%20underlying%20their%20language-specific%20processing%20remain%20largely%20unclear.%20In%20this%20paper%2C%20we%20investigate%20how%20LLMs%20handle%20multilingualism%20through%20the%20lens%20of%20sparse%20autoencoders%20%28SAEs%29%2C%20uncovering%20distinctive%20patterns%20that%20offer%20new%20insights%20into%20their%20internal%20workings.%20Specifically%2C%20we%20introduce%20two%20novel%20concepts%5Cu2014task%20instruction%5Cu2013focused%20%28TF%29%20and%20heading-focused%20%28HF%29%20SAE%20features%5Cu2014and%20use%20them%20to%20reveal%20intrinsic%20discrepancies%20between%20high-%20and%20low-performing%20languages.%20Our%20analysis%20yields%20several%20key%20findings%3A%20%281%29%20SAEs%20provide%20concrete%20evidence%20that%20LLMs%20have%20a%20precise%20understanding%20of%20prompt%20structure%3B%20%282%29%20heading%20keywords%20%28e.g.%2C%20%5Cu201cQuestion%2C%5Cu201d%20%5Cu201cChoices%2C%5Cu201d%20and%20%5Cu201cAnswer%5Cu201d%29%20play%20a%20distinct%20role%20in%20LLM%20processing%3B%20and%20%283%29%20low-performing%20languages%20exhibit%20a%20relative%20deficiency%20in%20TF%20features%20compared%20to%20high-performing%20languages.%20Building%20on%20these%20insights%2C%20we%20propose%20two%20practical%20strategies%20to%20improve%20zero-shot%20multilingual%20performance%3A%20%281%29%20incorporating%20English%20heading%20keywords%20and%20%282%29%20amplifying%20TF%20features%20through%20steering.%20Our%20approach%20improves%20zero-shot%20performance%20in%20low-performing%20languages%20by%20up%20to%203.7%25%20on%20average%20on%20ARC-Challenge%20and%20MMLU%2C%20while%20also%20shedding%20new%20light%20on%20fundamental%20differences%20between%20high-%20and%20low-performing%20languages%20in%20LLMs.%20Our%20code%20is%20available%20at%20https%3A%5C%2F%5C%2Fgithub.com%5C%2Fihcho2%5C%2FSAE-ML.%22%2C%22proceedingsTitle%22%3A%22%22%2C%22conferenceName%22%3A%22Second%20Conference%20on%20Language%20Modeling%22%2C%22date%22%3A%222025%5C%2F08%5C%2F26%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%22%22%2C%22ISBN%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fopenreview.net%5C%2Fforum%3Fid%3DNmGSvZoU3K%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222025-09-22T14%3A10%3A47Z%22%7D%7D%5D%7D
1.
Yu, H., Ahuja, K. K., Sankar, L. N. & Bryngelson, S. H. Transmission of High-Amplitude Sound through Leakages of Ill-fitting Earplugs. Preprint at https://doi.org/10.48550/ARXIV.2510.16355 (2025).
1.
Xiao, X. et al. Prompt-based Adaptation in Large-scale Vision Models: A Survey. Preprint at https://doi.org/10.48550/ARXIV.2510.13219 (2025).
1.
Maurel-Oujia, T. & Maeda, K. Subcontinuum structures of reactive shock waves in gaseous H 2 / O 2 mixtures. Phys. Rev. Fluids 10, 103201 (2025).
1.
Monea, G., Feldman, Y., Padmanabhan, S., Brantley, K. & Artzi, Y. Breadcrumbs Reasoning: Memory-Efficient Reasoning with Compression Beacons. Preprint at https://doi.org/10.48550/ARXIV.2510.13797 (2025).
1.
Xia, J. et al. Inpainting the Neural Picture: Inferring Unrecorded Brain Area Dynamics from Multi-Animal Datasets. Preprint at https://doi.org/10.48550/ARXIV.2510.11924 (2025).
1.
Ahmed, M. T., Ahmed, F. & Li, J. Enhancement of diffusivity and plastic deformation in ultrasound-assisted cold spray of tungsten: a molecular dynamics study. Preprint at https://doi.org/10.48550/ARXIV.2510.09803 (2025).
1.
Dhruv, V., Prather, B., Chandra, M., Joshi, A. V. & Gammie, C. F. Electromagnetic Observables of Weakly Collisional Black Hole Accretion. Preprint at https://doi.org/10.48550/ARXIV.2510.11365 (2025).
1.
Halychanskyi, Y., Churchwell, C., Wen, Y. & Kindratenko, V. FAC-FACodec: Controllable Zero-Shot Foreign Accent Conversion with Factorized Speech Codec. Preprint at https://doi.org/10.48550/ARXIV.2510.10785 (2025).
1.
Guan, H. & Balakrishna, A. R. CuPyMag: GPU-Accelerated Finite-Element Micromagnetics with Magnetostriction. Preprint at https://doi.org/10.48550/ARXIV.2510.09812 (2025).
1.
Liu, P., Rhee, S., Wilkins, M. & Dinda, P. Parameterized Algorithms and Parameter Selection for Fast GPU-GPU Collective Communication. Preprint at https://scholar.google.com/citations?view_op=view_citation&hl=en&user=wWZq-M0AAAAJ&citation_for_view=wWZq-M0AAAAJ:zYLM7Y9cAGgC.
1.
Hayes, T. L. & Krishnan, G. P. Quantifying the Role of OpenFold Components in Protein Structure Prediction. Preprint at https://doi.org/10.48550/ARXIV.2511.14781 (2025).
1.
Liu, L., Qin, F., Legunsen, O. & d’Amorim, M. Bug Histories as Sources of Compiler Fuzzing Mutators. Preprint at https://doi.org/10.48550/ARXIV.2510.07834 (2025).
1.
Alatrush, N. et al. Advancing Active Learning with Ensemble Strategies. in (Varna, Bulgaria, 2025).
1.
Arora, S. et al. Chain-of-Thought Reasoning in Streaming Full-Duplex End-to-End Spoken Dialogue Systems. Preprint at https://doi.org/10.48550/ARXIV.2510.02066 (2025).
1.
Yan, J. & Snir, M. Contemplating a Lightweight Communication Interface for Asynchronous Many-Task Systems. in Asynchronous Many-Task Systems and Applications (eds Diehl, P., Cao, Q., Herault, T. & Bosilca, G.) 165–173 (Springer Nature Switzerland, Cham, 2026). http://doi.org/10.1007/978-3-031-97196-9_14.
1.
Su, B.-H. et al. Reasoning Beyond Majority Vote: An Explainable SpeechLM Framework for Speech Emotion Recognition. Preprint at https://doi.org/10.48550/ARXIV.2509.24187 (2025).
1.
Habib, S. et al. Error quantification and comparison of binary neutron star gravitational waveforms from numerical relativity codes. Preprint at https://doi.org/10.48550/ARXIV.2509.23028 (2025).
1.
Zi, Y., Wu, Z., Boruch-Gruszecki, A., Bell, J. & Guha, A. AgentPack: A Dataset of Code Changes, Co-Authored by Agents and Humans. Preprint at https://doi.org/10.48550/ARXIV.2509.21891 (2025).
1.
Bhatnagar, S., Xu, A., Tan, K.-H. & Ahuja, N. LUQ: Layerwise Ultra-Low Bit Quantization for Multimodal Large Language Models. Preprint at https://doi.org/10.48550/ARXIV.2509.23729 (2025).
1.
Srivastava, G. et al. BeyondBench: Benchmark-Free Evaluation of Reasoning in Language Models. Preprint at https://doi.org/10.48550/ARXIV.2509.24210 (2025).
1.
Marx, E. et al. Likelihood-free inference for gravitational-wave data analysis and public alerts. Preprint at https://doi.org/10.48550/ARXIV.2509.22561 (2025).
1.
Liu, Z., Gong, A. & Clark, B. K. Decoding quantum low density parity check codes with diffusion. Preprint at https://doi.org/10.48550/arXiv.2509.22347 (2025).
1.
Halevi, G., Shankar, S., Mösta, P., Haas, R. & Schnetter, E. A Black Hole is Born: 3D GRMHD Simulation of Black Hole Formation from Core-Collapse. Preprint at https://doi.org/10.48550/arXiv.2506.20837 (2025).
1.
Lewis, A., Himberg, B., Torres-Sánchez, A. & Vanegas, J. Microscopic elasticity from MD part I: Bulk solid and fluid systems. Preprint at https://doi.org/10.26434/chemrxiv-2025-j17x4 (2025).
1.
Zhao, Y., Wang, Z. & Zhang, M. PuzzleMoE: Efficient Compression of Large Mixture-of-Experts Models via Sparse Expert Merging and Bit-packed inference. Preprint at https://doi.org/10.48550/ARXIV.2511.04805 (2025).
1.
Loehr, K. & Clark, B. K. Enhancing Neural Network Backflow. Preprint at https://doi.org/10.48550/ARXIV.2510.26906 (2025).
1.
Cheng, C.-H., Ficarra, G. & Witek, H. Dephasing in binary black hole mergers surrounded by scalar wave dark matter clouds. Preprint at https://doi.org/10.48550/ARXIV.2510.20037 (2025).
1.
Zhang, Z. A. et al. One Token per Highly Selective Frame: Towards Extreme Compression for Long Video Understanding. in (2025).
1.
Cui, S. et al. Story of Two GPUs: Characterizing the Resilience of Hopper H100 and Ampere A100 GPUs. in Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis 1145–1164 (ACM, St. Louis MO USA, 2025). http://doi.org/10.1145/3712285.3759821.
1.
Cross-Domain Long-Term Forecasting: Radiation Dose from Sparse Neutron Sensor via Spatio-Temporal Operator Network. https://arxiv.org/html/2510.18041v1.
1.
Chen, H. et al. ERA: Transforming VLMs into Embodied Agents via Embodied Prior Learning and Online Reinforcement Learning. Preprint at https://doi.org/10.48550/ARXIV.2510.12693 (2025).
1.
Wu, M. & Zhang, Z. Maple: A Multi-agent System for Portable Deep Learning across Clusters. Preprint at https://doi.org/10.48550/ARXIV.2510.08842 (2025).
1.
Xie, H. et al. Diamond: Harnessing GPU Resources for Scientific Deep Learning. in 2025 IEEE International Conference on eScience (eScience) 196–204 (IEEE, Chicago, IL, USA, 2025). http://doi.org/10.1109/eScience65000.2025.00031.
1.
Patel, P. et al. RADAR-Radio Afterglow Detection and AI-driven Response: A Federated Framework for Gravitational Wave Event Follow-Up. Preprint at https://doi.org/10.48550/ARXIV.2507.14827 (2025).
1.
Kacmaz, S., Haas, R. & Huerta, E. A. Machine Learning-Driven Conservative-to-Primitive Conversion in Hybrid Piecewise Polytropic and Tabulated Equations of State. Symmetry 17, 1409 (2025).
1.
Zhu, M. et al. Probing the Critical Point (CritPt) of AI Reasoning: a Frontier Physics Research Benchmark. Preprint at https://doi.org/10.48550/ARXIV.2509.26574 (2025).
1.
Lian, X., Tanaka, M., Ruwase, O. & Zhang, M. SuperOffload: Unleashing the Power of Large-Scale LLM Training on Superchips. Preprint at https://doi.org/10.48550/ARXIV.2509.21271 (2025).
1.
Gerogiannis, G. et al. NetSparse: In-Network Acceleration of Distributed Sparse Kernels. in 58th IEEE/ACM International Symposium on Microarchitecture (ACM, Seoul, Republic of Korea, 2025). doi:https://doi.org/10.1145/3725843.3756076.
1.
Wong, G. N., Medeiros, L. & Stone, J. M. Mass Transport, Turbulent Mixing, and Inflow in Black Hole Accretion. Preprint at https://doi.org/10.48550/ARXIV.2509.14202 (2025).
1.
Zhong, Y. & Most, E. R. Unraveling the emission mechanism powering long period radio transients from interacting white dwarf binaries via kinetic plasma simulations. Preprint at https://doi.org/10.48550/ARXIV.2509.09057 (2025).
1.
Agrawalla, B., Nauman, M., Agarwal, K. & Kumar, A. floq: Training Critics via Flow-Matching for Scaling Compute in Value-Based RL. Preprint at https://doi.org/10.48550/ARXIV.2509.06863 (2025).
1.
Jafari, S., Sarkar, A., Bilwal, M. & Jannesari, A. ProfilingAgent: Profiling-Guided Agentic Reasoning for Adaptive Model Optimization. Preprint at https://doi.org/10.48550/ARXIV.2509.05584 (2025).
1.
Wilfong, B. et al. Testing and benchmarking emerging supercomputers via the MFC flow solver. Preprint at https://doi.org/10.48550/ARXIV.2509.13575 (2025).
1.
Bazavov, A. et al. High-Precision Scale Setting with the Omega-Baryon Mass and Gradient Flow. Preprint at https://doi.org/10.48550/ARXIV.2509.14367 (2025).
1.
Yu, J., Taneja, A., Lin, J. & Zhang, M. VoltanaLLM: Feedback-Driven Frequency Control and State-Space Routing for Energy-Efficient LLM Serving. Preprint at https://doi.org/10.48550/ARXIV.2509.04827 (2025).
1.
Wang, S., Nguyen, H., Eick, B., Wilson, R. & El-Gohary, N. Efficient domain adaptation network for corrosion segmentation in infrastructure: from laboratory to real-world images. J Civil Struct Health Monit https://doi.org/10.1007/s13349-025-01003-y (2025) doi:10.1007/s13349-025-01003-y.
1.
Bhatia, M. et al. \textit{In Silico} Benchmarking of Detectable Byzantine Agreement in Noisy Quantum Networks. Preprint at https://doi.org/10.48550/ARXIV.2509.02629 (2025).
1.
Chantraine, C. et al. Ultrastrong Staphylococcus aureus adhesion to human skin: Calcium as a key regulator of noncovalent interactions. Sci. Adv. 11, eadu7457 (2025).
1.
Wang, P., Watanabe, S. & Van hamme, H. SSVD: Structured SVD for Parameter-Efficient Fine-Tuning and Benchmarking under Domain Shift in ASR. Preprint at https://doi.org/10.48550/ARXIV.2509.02830 (2025).
1.
Cho, I. & Hockenmaier, J. Analyzing Multilingualism in Large Language Models with Sparse Autoencoders. in (2025).